African linguistics
on the prairie
Selected papers from the 45th
Annual Conference on African
Linguistics
Edited by
Jason Kandybowicz
Travis Major
Harold Torrence
Philip T. Duncan
Contemporary African Linguistics 3
language
science
press
Contemporary African Linguistics
Editors: Akinbiyi Akinlabi, Laura J. Downing
In this series:
1. Payne, Doris L., Sara Pacchiarotti & Mokaya Bosire (eds.). Diversity in African languages:
Selected papers from the 46th Annual Conference on African Linguistics.
2. Persohn, Bastian. The verb in Nyakyusa: A focus on tense, aspect and modality.
3. Kandybowicz, Jason, Travis Major, Harold Torrence & Philip T. Duncan (eds.). African
linguistics on the prairie. Selected papers from the 45th Annual Conference on African
Linguistics.
ISSN: 2511-7726
African linguistics
on the prairie
Selected papers from the 45th
Annual Conference on African
Linguistics
Edited by
Jason Kandybowicz
Travis Major
Harold Torrence
Philip T. Duncan
language
science
press
Jason Kandybowicz, Travis Major, Harold Torrence & Philip T. Duncan (ed.). 2018.
African linguistics on the prairie: Selected papers from the 45th Annual Conference on
African Linguistics (Contemporary African Linguistics 3). Berlin: Language Science
Press.
This title can be downloaded at:
http://langsci-press.org/catalog/book/120
© 2018, the authors
Published under the Creative Commons Attribution 4.0 Licence (CC BY 4.0):
http://creativecommons.org/licenses/by/4.0/
ISBN: 978-3-96110-036-1 (Digital)
978-3-96110-037-8 (Hardcover)
ISSN: 2511-7726
DOI:10.5281/zenodo.1219141
Source code available from www.github.com/langsci/120
Collaborative reading: paperhive.org/documents/remote?type=langsci&id=120
Cover and concept of design: Ulrike Harbort
Typesetting: Birgit Jänen, Felix Kopecky, Iana Stefanova, Phil Duncan, Sebastian
Nordhoff
Proofreading: Alexis Michaud, Amr Zawawy, Andreas Hölzl, Bev Erasmus, Brett
Reynolds, Christian Döhler, Eitan Grossman, Elizabeth Bogal Allbritten, Evans Gesure,
Gerald Delahunty, Jean Nitzke, Jonathan Brindle, Ka Yau Lai, Ken Manson, M Lotta
Aunio, Maria Isabel Maldonado, Martin Haspelmath, Paulson Skerrit, Richard Griscom,
Rosey Billington, Steven Kaye
Fonts: Linux Libertine, Libertinus Math, Arimo, DejaVu Sans Mono
Typesetting software: XƎLATEX
Language Science Press
Unter den Linden 6
10099 Berlin, Germany
langsci-press.org
Storage and cataloguing done by FU Berlin
Contents
Preface
v
I
General linguistics
1
Linguistic complexity: A case study from Swahili
Kyle Jerro
2
Inter-party insults in political discourse in Ghana: A critical discourse
analysis
Emmanuel Amo Ofori
21
Classification of Guébie within Kru
Hannah Leigh Sande
37
3
4 What about Southern African story grammar? Promoting language specific
macrostructures in educational settings
Heike Tappe
5
How multilingual policies can fail: Language politics among Ethiopian
political parties
Mehari Zemelak Worku
3
51
65
II Language endangerment
6
7
Linguistic imperialism and language decolonisation in Africa through
documentation and preservation
Kofi Agyekum
Dictionary Day: A community-driven approach to dictionary compilation
Bryan D. Gelles
8 Language endangerment in Southwestern Burkina: A tale of two Tiefos
Abbie Hantgan-Sonko
87
105
117
Contents
III Morphology and phonology
9 Consonant substitution in child language (Ikwere)
Roseline I. C. Alerechi
135
10 A morphosyntactic analysis of adjectives in two Kwa languages: Ga and
Dangme
Regina Oforiwah Caesar & Yvonne A. A. Ollennu
157
11 Towards a unified theory of morphological productivity in the Bantu
languages: A corpus analysis of nominalization patterns in Swahili
Nick Kloehn
175
12 The acoustic vowel space of Anyi in light of the cardinal vowel system and
the Dispersion Focalization Theory
Ettien Koffi
191
13 Gender instability in Maay
Mary Paster
205
14 Egyptian Arabic broken plurals in DATR
Lindley Winchester
219
IV Syntax
15 Factive relative clauses in Pulaar
Ibrahima Ba
235
16 Object suffixes as incorporated pronouns in Seereer
Nico Baier
253
17 Searching high and low for focus in Ibibio
Philip T. Duncan, Travis Major & Mfon Udoinyang
269
18 More on have and need
Claire Halpert & Michael Diercks
289
19 Structural transfer in third language acquisition: The case of Lingala-French
speakers acquiring English
Philothé Mwamba Kabasele
307
20 Adjectives in Lubukusu
Aggrey Wasike
ii
325
Contents
21 Optional ergativity and information structure in Beria
Andrew Wolfe & Tajeldin Abdalla Adam
341
V Semantics and pragmatics
22 Two-place exceed comparatives in Luganda
M. Ryan Bochnak
361
23 Temporal remoteness and vagueness in past time reference in Luganda
M. Ryan Bochnak & Peter Klecha
377
24 Focus marking in Kuria
Meredith Landman & Rodrigo Ranero
393
25 A corpus study of the Swahili demonstrative position
Mohamed Mwamzandi
413
Index
431
iii
Preface
African Linguistics on the Prairie features select peer-reviewed papers from the 45th Annual Conference on African Linguistics (ACAL 45). The conference was held on April
17-19, 2014 at the University of Kansas in Lawrence, Kansas and was hosted by the Department of Linguistics. The plenary speakers for ACAL 45 were: Kofi Agyekum, University
of Ghana; Chris Collins, New York University; Ruth Kramer, Georgetown University;
Michael R. Marlo, University of Missouri; Carlos M Nash, University of California, Santa
Barbara; Bonny Sands, Northern Arizona University; and Malte Zimmermann, Universität Potsdam. The theme of the conference was “Africa’s Endangered Languages: Documentary and Theoretical Approaches”. In conjunction with the conference, a special
three-day workshop, supported by a grant from the National Science Foundation (NSF
DEL 1360823), was organized around the conference theme. The workshop brought together scholars with varied perspectives and research agendas to address the unique
challenges facing endangered languages, language documentation, and revitalization efforts in Africa. Select papers from that workshop appear in the volume Africa’s Endangered Languages, published in 2017 by Oxford University Press. The articles appearing in
this volume were for the most part presented in the main session of the conference.
The articles that comprise this volume reflect the enormous diversity of African languages, as they focus on varieties from all of the major African language phyla. The
articles here also reflect the many different research perspectives that frame the work
of linguists in the Association for Contemporary African Linguistics. The diversity of
views presented here are thus indicative of the vitality of current African linguistics research. As a perusal of the titles hints, the work published in this volume covers fields
ranging from phonetics, phonology, morphology, typology, syntax, and semantics to
sociolinguistics, discourse, analysis, language acquisition, computational linguistics and
beyond. This broad scope and the quality of the articles contained within holds out the
promise of continued advancement in linguistic research on African languages.
ACAL 45 would not have been possible without financial support from multiple institutions. We gratefully acknowledge support from the University of Kansas Department of Linguistics, the Kansas African Studies Center, the office of the Dean of the
College of Liberal Arts and Sciences, and the National Science Foundation. The ACAL
45 Organizing Committee, Travis Major, Ibrahima Ba, Mfon Udoinyang, Carlos Nash
and Peter Ojiambo, played a central role in the success of the conference and we thank
them immensely for their tireless work and support. The University of Kansas Linguistics Department provided more than just material support. Department faculty, graduate
students, and staff volunteered hours of their time at all stages of the planning and execution of the conference. We thank: Saad Aldossari, Chia-Ying Chu, Katrina Connell,
Preface
Kate Coughlin, Lauren Covey, Philip Duncan, Robert Fiorentino, Alison Gabriele, Longcan Huang, Corinna Johnson, Allard Jongman, David Kummer, Mingxing Li, Beatriz
Lopez Prego, María Martínez García, Andrew McKenzie, Utako Minai, Zhen Qin, Maria
Rangel, Sara Rosen, Joan Sereno, Khady Tamba, Wenting Tang, Annie Tremblay, Xiao
Yang, and Jie Zhang. The following individuals served as session chairs at the conference and we would like to express our thanks to them as well: Akin Akinlabi, Andrew
McKenzie, Mike Cahill, Jeanine Ntihirageza, Vicki Carstens, Allard Jongman, Lindley
Winchester, Martha Michieka, Ibrahima Ba, Timothy M. Stirtz, Christopher Green, Annie Tremblay, Jie Zhang, Kasangati Kinyalolo, Lisa Zsiga, Rebecca Hale, Peter Ojiambo,
Claire Halpert, Lee Bickmore, Mohamed Mwamzandi, Heike Tappe, Laura McPherson,
Tucker Childs, Philip Rudd, Mary Paster, Peter Jenks, Rose-Marie Déchaine, Michael
Diercks, Olanike Orie, Mamadou Bassene, and James Essegbey. In putting together this
volume, we have relied upon the many specialists who generously agreed to serve as reviewers, thus ensuring the high quality of articles that appear within. We are exceedingly
grateful to: Oluseye Adesola, Colleen Ahland, Assibi Amidu, Anton Antonov, Rebekah
Baglini, Nicholas Baier, Anna Belew, Kelly Harper Berkson, Leston Buell, Tucker Childs,
Caitlin Coughlin, Rose-Marie Déchaine, Katherine Demuth, Michael Diercks, Philip Duncan, John Gluckman, Scott Grimm, Claire Halpert, Claire Harter, Stefanie Harves, Peter
Jenks, Gregory Kobele, Ruth Kramer, Jorge Emilio Rosés Labrada, Florian Lionnet, Victor
Manfredi, Michael Marlo, Andrew McKenzie, Laura McPherson, Leonard Muaka, Samuel
Gyasi Obeng, Doris Payne, Clifton Pye, Philip Rudd, Russell Schuh, Anne Schwarz, Peter
Trudgill, Matthew Tucker, Jenneke van der Wal, and Malte Zimmermann. Thanks also
to Joan Maling at the National Science Foundation and Kate Lorenz at the University of
Kansas Institute for Policy and Social Research for valuable assistance in helping us navigate through the complexities of external funding. We also acknowledge the valuable
assistance provided by Akin Akinlabi and Lee Bickmore, who not only answered countless questions and offered excellent advice, but helped run the ACAL organization and
created the infrastructure for the publication of this and future ACAL volumes with Language Science Press. Lastly, we heartily thank Sebastian Nordhoff at Language Science
Press for his invaluable help with the many technical aspects of publishing this volume.
vi
Part I
General linguistics
Chapter 1
Linguistic complexity: A case study from
Swahili
Kyle Jerro
University of Essex
This paper addresses the question of linguistic complexity in Swahili, a Bantu language spoken in East and Central Africa. Literature on linguistic complexity in other languages has
argued that high levels of second-language learning affect linguistic complexity over time.
Swahili serves as an ideal case study for this question because it has been used as a lingua franca for several centuries. I compare the phonological and morphological systems in
Swahili to five other related Bantu languages, as well as compare all six languages to the
original Proto-Bantu systems. The results of the study show that there is no decrease in
phonological or morphological complexity in (standard) Swahili when compared to other
closely related Bantu languages, though the grammar has strongly diverged from the other
related languages.
1 Introduction: the question of linguistic complexity
It is generally assumed by linguists that all languages share the same level of complexity, with “simpler” areas of grammar being compensated by more complexity elsewhere.
Some researchers take this as a core design feature of language (cf. work from the generative perspective, such as Pinker & Bloom 1990; Pinker 1994; Baker 2003), though this
has tacitly pervaded most linguistic thought.
Recently, however, work by various linguistic typologists has put this assumption into
question, investigating several linguistic domains (see Miestamo 2008; Sampson 2009
and Givón & Shibatani 2009 for overviews of the literature on complexity). A core area
of the research in this field is simply how to answer such a question (Nichols 2009;
Sampson 2009; Miestamo et al. 2008). For example, Nichols (2009) compares various
features of languages, such size of phoneme inventory, number of inflectional categories
on a basic verb, number of alignments in a single language, etc. Other work situates
linguistic complexity within a social context. One claim is that older languages tend to
be more complex that new ones (e.g. Creoles), cf. McWhorter (2008) and Trudgill (2009).
Kyle Jerro. Linguistic complexity: A case study from Swahili. In Jason Kandybowicz, Travis
Major, Harold Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected
papers from the 45th Annual Conference on African Linguistics, 3–19. Berlin: Language Science
Press. DOI:10.5281/zenodo.1251708
Kyle Jerro
Another claim is that population size relates to linguistic complexity (Trudgill 2004; Hay
& Bauer 2007; Nichols 2009).
Another vein of this literature – and the topic of this paper – has investigated the
interaction of complexity and language contact, claiming that high amounts of secondlanguage learning, including the use as a lingua franca, affects linguistic complexity and
increases the rate of language change (Kusters 2003a,b; Trudgill 2009; McWhorter 2008;
2011; Trudgill 2011). Trudgill (2011) claims that that the specific effect on complexity is
contingent upon the nature of second-language learning: while large amounts of secondlanguage learning by adult speakers may result in net decomplexification, learning by
children (e.g. through prolonged contact between two languages) may lead to increased
complexity. This paper tests the affects of language contact on complexity in Swahili,
used as a lingua franca throughout much of East and Central Africa. I compare Standard
Swahili to neighboring Bantu languages in their synchronic morphological and phonological features as well as their divergence from Proto-Bantu.
To test this claim, I employ similar metrics of complexity to those used by Kusters and
McWhorter (i.e morphology, see §5), comparing different aspects of Swahili morphology
to the grammar of five sister languages. In addition, I discuss the phonological inventories of the languages, a component absent from Kusters’ and McWhorter’s studies, but
discussed at length by others (Hay & Bauer 2007; Trudgill 2011). From the comparisons, I
conclude that Swahili does not exhibit any systematic decomplexification in comparison
to the other languages, though it shows several grammatical differences from related languages. This situation is predicted from the framework proposed in Trudgill (2011), where
long-term bilingualism (here, between Swahili and Arabic) may lead to the rapid change
of a contact language.
The remainder of this paper is organized as follows: in §2, I summarize the claims of the
decomplexification hypothesis. I then outline the linguistic and sociolinguistic situations
of five Bantu languages from East Africa chosen to serve as comparison cases. Sections
4-5 use phonological and morphological metrics, respectively, in order to compare the
complexity of Swahili to the comparison languages. Section 6 discusses the findings and
their relation to the the decomplexification hypothesis.
2 Contact and (de-)complexification
In research on complexity, two opposite effects on complexity have been found, depending on the nature of the linguistic community. Languages in prolonged contact regions tend to develop high amounts of linguistic complexity (Heine & Kuteva 2005; Dahl
2004; Givón 1984). On the other hand, situations with high numbers of sudden secondlanguage learners result in simplification of linguistic structure. As discussed in Trudgill
(2011), the crucial divide between the two groups is the critical period of language acquisition: adult learners are not as adept as children at acquiring a (second) language.
In a situation where adult speakers are acquiring a language, this “sub-optimal acquisition” (a term from Dahl 2004) results in the reduction of ornamental or non-obligatory
elements of grammar.
4
1 Linguistic complexity: A case study from Swahili
As Kusters (2003b) states, “the more second-language learning has taken place in a
speech community, the more internal dialect contact and migrations occurred, and the
less prestige a language has, the more transparent and economic the verbal inflection will
become” (275, emphasis in original). For Kusters, an inflectional system is more economic
if it makes fewer category distinctions. In order to test the prediction of the decomplexification hypothesis, lingua francas that have been used by many second-language learners
can be compared to sister languages or varieties that have not been used as lingua francas.
Kusters (2003b,a) provides several case studies in contact languages that have undergone decomplexification, tracing the changes from an older stage of the language to various modern sister languages. For example, one case study comes from three descendants
of Old Norse: Icelandic, Faroese, and Standard Norwegian. He argues that the varieties
that are more insular have maintained complexity that is absent in metropolitan varieties (i.e. the dialect of the capital city of the Faroese Islands, Tórshavn). As an example,
consider the data in Table 1, with the verb forms for the verb ‘to awake’ in Old Norse
and three descendant languages (Kusters 2003b: 285, Table 5).
Table 1: Verbal tense in Old Norse and descendant languages
1sg
2sg
3sg
1pl
2pl
3pl
Old Norse
Icelandic
Faroese
Tórshavn
vakn-a
vakn-ar
vakn-ar
vakn-um
vakn- i_
vakn-a
vakn-a
vakn-ar
vakn-ar
vökn-um
vakn-ið
vakn-a
vakn-i
vakn-ar
vakn-ar
vakn-a
vakn-a
vakn-a
(-′ )
(-′ r)
(-′ r)
(-′ )
(-′ )
(-′ )
He argues that Faroese, a variant that has been in prolonged contact with Danish, has
reduced morphological complexity from the Old Norse, and Tórshavn has undergone further reduction, having only stress as a indicator of tense. The only person marking is the
marking of second- and third-singular, to the exclusion of all other persons and numbers. In addition, the Tórshavn dialect has completely neutralized certain inflectional
categories, like past indicative and present subjunctive.
McWhorter (2011; 2008) makes the stronger claim that second-language learning is
the only factor that drives overall simplification in a language. Namely, sweeping loss of
complexity in a language is impossible without the influence of second-language learning. The argument works in the opposite direction from Kusters’; when you find an
instance of decomplexification, it is predicted that this must have come from a situation of high second-language learning. McWhorter’s metrics of complexity are similar
to those of Kusters (2003b). For example, in his 2008 paper, he compares two varieties of
the Tetun language spoken in Timor. The first, Tetun Dili, is used as a lingua franca by
5
Kyle Jerro
two-thirds of the island; the other, Tetun Terik, is only spoken on the southern coastline.
McWhorter predicts that because Tetun Dili is a lingua franca, it has a simpler grammar
than Tetun Terik. He presents several instances where the Dili variety is more economical in the number of morphological categories it has. For example, while Terik has three
verbal affixes, Dili has two; Tetun has six numeral classifiers while Dili only has four
(and those four are used optionally); Tetun has an overt marker for definiteness, while
Dili uses context to indicate this; Tetun has three copulas, while Dili has only one; etc.
In short, the variety that is used as a lingua franca is systematically simpler than a sister
variety without the same level of second-language use.
When two languages are in prolonged contact, and the acquirers of a second language
are mostly children, the opposite effect is found: over time, more complexity is found,
often by the additive borrowing from the neighboring language. For example, Comrie
(2008) and Trudgill (2011) cite the example of Michif, a mixed language from contact
between Cree and French (Bakker 1997). Michif, from prolonged multilingualism with
French and Cree, developed an elaborate grammar, taking grammatical elements from
both Cree and French, with verbal structure inherited from the former and nominal structure from the latter. The result is that Michif employs elaborate verbal and morphological
categories found in neither French nor Cree.
In short, work on contact and complexity has found three related effects of contact:
first, language contact increases the rate of language change; second, second-language
learning by adults often leads to reduction in complexity via imperfect acquisition; and,
third, prolonged contact between two languages often results in complexification as
forms are taken from one and added into the other. In this paper, I tease apart the level of
complexity of standard Swahili, comparing it to five related Bantu languages that have
not had parallel situations of language contact.
3 Swahili and the five comparison languages
Swahili serves as another ideal case study in fleshing out the claims of the decomplexification hypothesis. Swahili is spoken as a native language along the Indian Ocean coast
of Kenya and Tanzania and in the Zanzibari archipelago. It is also used as an official language and lingua franca in Kenya, Tanzania and the Democratic Republic of the Congo
(DRC) in addition to a language of business and commerce at different points in history
in Uganda, Rwanda, and Burundi. Because of this widespread use as a lingua franca,
nearly 140 million people use Swahili as a second language, while only 5 million speak it
natively. Given the overwhelming predominance of second-language speakers of the language, the decomplexification hypothesis predicts that Swahili should be systematically
less complex than related languages with little or no use by second-language speakers.
I have chosen five languages spoken in the countries where Swahili is or has been
routinely used as a lingua franca. I have chosen one language from each country, and the
languages are all part of the Northeastern branch of the Bantu family (with the exception
6
1 Linguistic complexity: A case study from Swahili
of Lingala).1 The comparison languages are Gikuyu (Kenya, E.51), Lingala (DRC, C.30B),
Haya (Tanzania, JE.22), Kinyarwanda (Rwanda, DJ.61), and Luganda (Uganda, JE.15).
Gikuyu is spoken in Central Kenya by the Gikuyu people, numbering at approximately
7 million. Lingala is a language spoken by approximately 2 million people in the Republic
of Congo, the Democratic Republic of Congo, and parts of the Central African Republic. Haya is spoken in Northwestern Tanzania, near the shores of Lake Victoria (Byarushengo et al. 1977). There are approximately 1 million speakers of the language. Luganda
is spoken by approximately 4 million people in Southern Uganda. Though used mostly
by the Baganda people, it is also used as a second language by approximately 1 million
people in Uganda (Ethnologue 2013). Although the use of Luganda by second-language
learners is not ideal as a comparison case in the current study, the situation of Luganda
is different from Swahili in that the majority of speakers use Luganda as a first language.
Swahili on the other hand, is used overwhelmingly as a second language. Kinyarwanda
is spoken by somewhere around 12 million people in Rwanda, Burundi, and parts of
Uganda and DRC.
4 Phonological complexity
The first metric I use to compare the relative complexity among these languages is
their phonological inventories. Phonological complexity did not figure in Kusters’ and
McWhorter’s discussions, though several other works have used phonological inventory
as a metric for calculating complexity (Hay & Bauer 2007; Nichols 2009). The decomplexification hypothesis as outlined above predicts that Swahili will have the smallest inventory of phonemes; over time, imperfect learning by second-language speakers would
result in the reduction of phoneme contrasts not found in their first languages. Over
time, this reduced vowel inventory becomes the standard inventory of the language.
4.1 Vowel complexity
4.1.1 Vowel inventory
Bantu languages generally have between five and seven vowels in their inventory, and
they generally include tonal and length distinctions (Hyman 2003; Maddieson 2003).
Proto-Bantu has been reconstructed to have seven vowels with high and low tone contrasts. Table 2 indicates the number of different vowels (based on quality) in each of the
languages in the test set as well as whether each language makes a distinction between
long and short vowels and between tones.
Numerically, Swahili has a simpler vowel inventory than the other languages; it has
two fewer vowels than Proto-Bantu. Furthermore, Swahili has lost the tone and length
1
A better comparison set may be languages that are more closely related to Swahili genetically than the
five chosen here. Accessibility to resources was a major factor in linguistic choice, though the localization
of these languages to East Africa is intentionally aimed at keeping to languages that are more similar to
Swahili.
7
Kyle Jerro
Table 2: Size of vowel inventories
Language
Vowels
Tone
Length
Source
Proto-Bantu
Swahili
Gikyuyu
Lingala
Haya
Kinyarwanda
Luganda
7
5
7
7
5
5
5
+
–
+
+
+
+
+
+
–
+
+
+
+
+
Maddieson (2003)
Ashton (1966)
Barlow (1960)
Guthrie (1966)
Byarushengo et al. (1977)
Myers & Crowhurst (2006)
Kirwan & Gore (1951)
contrasts in Proto-Bantu, while the other languages have retained these features. This is
the kind of inventory reduction expected by the decomplexification hypothesis.
4.1.2 Other kinds of vowel complexity
Although the size of vowel inventories indicates a lower level of complexity in Swahili,
another possible metric is linguistic markedness (cf. McWhorter 2008; 2011). Swahili, unlike its sister languages, shows three linguistically marked phonological processes that
are absent in the other languages. These processes include the permission of syllabic
consonants, an irregular stress system, and vowel hiatus. Unlike a numerical metric like
phoneme inventory, however, phonological operations in a language are not as easily
quantifiable. However, I argue here that the quantitatively fewer phonemic vowel contrasts in Swahili are counteracted by the complexity that ensues with respect to its vowel
system.
First, Swahili has syllabic nasal consonants (Ashton 1966). This is present on words
such as mtoto [m.toto] ‘child,’ mtu [m.tu] ‘person,’ and mlango [m.lango] ‘door.’ Of the
sister languages, only Haya permits syllabic consonants; all maintain a minimal (C)CV
syllable structure (cf. the cited grammars). Interestingly, Hyman (2003) assumes this is
a natural change, derived from the loss of [u] in mu- nominal prefixes.
A further noteworthy difference between Swahili to the exclusion of the other languages is that Swahili permits vowel hiatus, with juxtaposed vowels serving as nuclei
of separate syllables. For example, chui ‘leopard’ is syllabified as [tʃu.i], and paa ‘gazelle’
as [pa.a]. The other languages do not permit vowel hiatus; Kinyarwanda, for example,
deletes one of any two adjacent vowels, even between word boundaries. For example, the
sentence uri umwana ‘you are a child’ is pronounced [u.ru.mŋa.na], with the word-final
[i] in uri being deleted.
Finally, unlike the other languages of the study, Swahili has several cases of irregular lexical stress.2 In most Bantu languages, stress falls on the penultimate syllable. In
Swahili, however, there are cases where Arabic loanwords carry stress on the antepenul2 Thanks
8
to Scott Myers for suggesting this point.
1 Linguistic complexity: A case study from Swahili
timate syllable, in words such as nusura [ˈnu.su.ra] ‘almost,’ ratili [ˈra.ti.li] ‘pound,’ and
thumuni [ˈth u.mu.ni] ‘an eighth’ (Ashton 1966). Here, contact with Arabic is the obvious
influence of the complexification of the Swahili stress system.
These three examples show that despite the smaller phonemic inventory, Swahili has
elements of complexity that are absent in the other languages. These features, however,
are difficult to quantify, and their inclusion in metrics of complexity vary. My conclusion
from the data in this section is that there is no clear reduction in complexity in the vowel
system of Swahili.
4.2 Consonant inventory
Although the number of vowels in Swahili is quite low, the consonant inventory is noticeably larger than the inventories of the comparison languages.3
Table 3: Size of consonant inventories
Language
Consonants
Proto-Bantu
Swahili
Gikuyu
Lingala
Haya
Kinyarwanda
Luganda
11
30
14
15
19
22
18
The consonant inventory in Swahili is striking larger than the other languages under
discussion, being over two times larger than the consonant inventory of Gikuyu and
Proto-Bantu.4 The larger inventory in Swahili comes in part from having both voiced
and voiceless stops and fricatives for bilabial, alveolar, and velar places of articulation.
Many languages lack a subset of these sounds, often having only the voiced or voiceless
counterpart. Gikuyu, for example, lacks the voiceless bilabial stop, the voiceless velar
fricative, and the voiced alveolar fricative that are found in Swahili.
A further difference is that Swahili is the only language in the group with the aspirated stops and fricatives [ ph th tʃh kh ] (Ashton 1966; Engstrand & Lodhi 1985). Aspiration is also found in various other Bantu languages, such as Zulu, Swati, Makua,
Doko, Chicheŵa, and Kongo. It has been argued that aspiration is a possible outgrowth
3
The inventories in Table 3 come from the same sources as in Table 2, save for the number for Proto-Bantu,
which comes from Hyman (2003).
4
Nasalized consonants were not counted for any of the languages, as the descriptions of them were not
satisfactorily convincing that these were indeed separate phonemes. The inclusion of these sounds in the
data would not affect the trend, however, since they are also a class of sounds reported in Swahili.
9
Kyle Jerro
of a consonant followed by the Proto-Bantu high vowels (Hyman 2003) or from an earlier voiceless pre nasalized stop (Maddieson 2003). Regardless of the origin of phonemic
aspiration, the presence of aspiration results in a notable increase in the phonemic inventory of Swahili, resulting in a larger inventory than the comparison languages, as
well as an innovation since Proto-Bantu.
Another interesting feature of the Swahili consonant system is that all voiced stops are
implosives. Swahili has four of these phonemes: [ ɓ ɗ ʄ ɠ ]. Implosive stops are not found
in any of the comparison languages from East Africa, though implosive stops are documented in the southern Bantu languages, with Maddieson (2003) treating implosives in
the Bantu family as a natural development in some daughter languages.
4.3 Discussion
The decomplexification hypothesis predicts that Swahili should have a noticeably smaller
phoneme inventory than the comparison languages. Although this is true with vowel inventory, the consonant inventory in Swahili is markedly larger than any of the other
comparison languages. Importantly, the Swahili consonant system is nearly three times
larger than in Proto-Bantu, suggesting considerable innovation during the evolution of
Swahili.
5 Morphological complexity
The next domain of investigation is the morphological (dis)similarity between Swahili
and the other Bantu languages. If the decomplexification hypothesis is correct, it is
expected that Swahili will make fewer distinctions and that morphemes will be more
phonologically reduced than the other languages. I investigate the domains of noun
class morphology, valency-changing morphology, and tense/aspect/mood morphology,
which are all three morphological domains that are found in each of the languages.
5.1 Gender classes on nominals
Bantu languages are well known for their rich noun class morphology. The noun classes
for Swahili, Haya, Kinyarwanda, Luganda, and Lingala are provided in Table 4, as well as
the reconstructions of the Proto-Bantu inventory (Meeussen 1967; Schadeberg 2003a).5
Given then decomplexification hypothesis, it is expected that Swahili should be more
economic in its morphological forms, either in the phonological shape of the morphemes
or in the number of semantic distinctions.
Swahili has a comparable number of category distinctions to the other languages; although it is reduced from Proto-Bantu, only one of the other languages retains the number of category distinctions found in Proto-Bantu (i.e. Luganda). Clearly, the prediction
5
10
The source for Gikuyu did not include enough detail for this comparison. The sources for the modern languages in Table 4 are: Swahili (Ashton 1966), Haya (Byarushengo et al. 1977), Kinyarwanda (kinyarwanda.net), Luganda (Kirwan & Gore 1951), and Lingala (Guthrie & Carrington 1988).
1 Linguistic complexity: A case study from Swahili
Table 4: Comparison of noun class morphology
Class
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Swahili
Haya
Kinyarwanda
Luganda
Lingala
PB
m(u)wam(u)mijimakivinnunkupakumu-
mubamumilimakibinnlukatubuku-
umuabaumuimiiriamaikiibii(n)i(n)uruakautuubuukuaha-
(o)mu(a)ba(o)mu(e)mili-, eri(a)ma(e)ki(e)bi(e)n(e)n(o)lu(a)ka(o)tu(o)bu(o)kuwakumu(o)gu(a)gae-
mobamomilimaebiNNloboboko-
*mu*ba*mu*mi*i̹*ma*ki*bi̹*n*n*du*ka*tu*bu*ku*pa*ku*mu*pi̹*i-
16
15
16
21
14
21
that Swahili exhibit a noteworthy reduction in the number category distinctions is not
borne out in this comparison.
As for the phonological shape of the morphemes, Swahili lacks the pre-prefix that is
found in Luganda and Rwanda. At a first glance, this could be argued to be an instance
of phonological reduction in Swahili. However, it has been argued in the literature that
these pre-prefixes were not present in Proto-Bantu (Katamba 2003), suggesting that the
pre-prefix in languages that have it is an innovation.
Support for this point is that the use of the pre-prefix varies drastically in the languages which use it. In Luganda, a variety of features converge to predict the presence
of the pre-prefix, such as whether the noun is a dependent or main clause, appears in
the affirmative or negative, etc. (Hyman & Katamba 1991; 1993). In Zulu, it has been argued that the pre prefix is a case marker for nominals that lack structural case (Halpert
2012). Zerbian & Krifka (2008) show that features such as genericity, specificity, and def-
11
Kyle Jerro
initeness are present in various languages which utilize the pre-prefix, such as Xhosa,
Bemba, and Kinande. Crucially, it is assumed that the pre-prefix is a later innovation
from Proto-Bantu, perhaps being a reanalysis of cliticized pronouns onto the main noun
(Bleek 1869).
The lack of a pre-prefix in the Proto-Bantu stems, as well as the semantic nature of preprefixes in the languages which have them, suggests that the reduced phonological shape
of class morphology in Swahili is not driven by phonological reduction due to secondlanguage learning. Instead, Swahili has retained the original shape of Proto-Bantu stems.
5.2 Valency-changing morphology
Bantu languages utilize morphology to indicate valency changes to the argument structure of a verb. Both argument-adding (applicatives and causatives) and argument-reducing (stative, reciprocal, passive) morphology is employed by these languages. If the decomplexification hypothesis is correct, it is expected that valency-changing morphology
in Swahili is simpler than in the comparison languages – be it phonologically reduced
or with fewer morphological category distinctions.
Table 5 gives the morphological forms for different valency-changing morphology in
Swahili (Russell 2003), Lingala (Guthrie 1966), Kinyarwanda6 (Jerro 2015), Haya (Byarushengo et al. 1977), and the reconstructed forms in Proto-Bantu (Schadeberg 2003b).7
Table 5: Comparison of valency-changing morphology
Type
Swahili
Lingala
Kinyarwanda
Haya
PB
Benefactive
Instrumental
Locative
Causative
-(l)e /-(l)i
-(l)e /-(l)i
-(l)e /-(l)i
-ish/-esh
-el
-is
-ir/-er
-ish/-esh
-ir/-er
-ish/-esh
-il/-el
-is/-es
-il/-el
-is/-es
*-ɪl
*–ɪl
*-ɪl
*-i/-ici
Stative
Reciprocal
Passive
-ik/-ek
-an
-(li)w/-(le)w
-an
-an
-
-ik/-ek
-an
-w
-ek
-aŋɡan
-w
*-ɪk
*-an
*-ʊ/-ɪbʊ
The first three types of morphology are applicatives, which add a new object to the
valency of a verb. Reciprocals, statives, and passives all decrease the valency of a verb by
one: reciprocals link the action back to the subject, i.e. the subject does the action to him
or herself; passives demote the subject to an oblique position and promote the object to
subject position; and statives describe the result state of a transitive verb.
6
Those familiar with Kimenyi (1980) will notice that the locative applicative morpheme for Kinyarwanda in
Table 5 differs from Kimenyi’s description. Jerro (2015) describes a different locative applicative form for
his speakers, who find Kimenyi’s locative applicatives ungrammatical.
7 The resources for Gikuyu and Luganda do not explicitly discuss valency-changing morphology.
12
1 Linguistic complexity: A case study from Swahili
Contrary to the decomplexification hypothesis, the data in Table 5 show that Swahili
does not have a simpler system of valency-changing morphology. From the perspective
of the number of category distinctions, it has a comparable number to the other languages, and has lost no form reconstructed for Proto-Bantu.
From the perspective of the phonological shape of the morphemes, there is no evidence that Swahili is simpler than the other languages. Many of the valency-changing
forms in Bantu undergo vowel harmony with the nearest stem on the vowel, and Swahili
is not an exception to this; it employs vowel harmony on valency-changing morphology
in the same way as its sister languages.
In fact, if any argument were to be made regarding the complexity of valency-changing
morphemes, Swahili is more complex in the phonological shape of its passive morpheme,
which varies by context depending on the phonological shape of the verb to which it is
applied (Russell 2003). The most productive form of the Swahili passive is –w, as in fung–
w–a from funga ‘fasten’ and tumi–w–a from tumia ‘use.’ When the verb stem ends in [o]
or [e], the form –lew is used. If the verb stem ends in [a] or [u], the form –liw is used, as
in za–liw–a from zaa ‘give birth’ and fu–liw–a from fua ‘wash clothes’. Russell (2003)
also notes that the passive forms –ew and –iw are used with verbs of Arabic origin, such
as sameh–ew–a from samehe ‘forgive’ and hitaj–iw–a from hitaji ‘need.’ In short, to form
a passive in Swahili, there are complex factors that determine the phonological shape of
the passive morpheme, and these factors are not present in the comparison languages.
In Kinyarwanda and Haya, for example, the passive form is –w for all verbs, and Lingala lacks a separate passive morpheme altogether (Guthrie 1966). This is evidence that
valency-changing morphology in Swahili is not simpler than the sister languages, and
in the domain of the passive, Swahili is actually more complex than the other forms.
5.3 Tense, aspect, and mood
Bantu languages have rich systems of tense, aspect, and mood (TAM). From the view
of complexity, there are two ways in which a language may be simpler than the others
with respect to TAM morphology. The language could make fewer distinctions in its
tense, aspect, and mood categories, leaving TAM information to pragmatics. Another
indication of decomplexification is if the language shows phonological reduction of the
forms compared to other languages or from the protolanguage.
In Bantu languages, aspect and mood morphology generally appears as a prefix before
the verb stem, but after the agreement subject marker. Aspect, on the other hand, appears
as a suffix after the verb stem. If a language marks subjunctive or indicative, this appears
in the aspect slot. The general template for TAM on a verb in Bantu is given in (1) (cf.
Meeussen 1967; Nurse 2003).
(1)
Subject Marker – Tense – stem – Aspect/Subjunctive
Table 6 includes data for five different kinds of TAM that are prevalent in Bantu languages: tense, indicative/subjunctive, aspect, negation, and idiosyncratic TAM morphology that does not fit consistently with the other categories.8
8
Data from Lingala and Haya are not included due to a paucity of description of the tense/aspect systems
in those languages.
13
Kyle Jerro
Table 6: Comparison of tense, aspect, and mood morphology
Type
Swahili
Luganda
Gikuyu
Rwanda
naalimeta-
∅
ams
naali-
∅
ra∅
a-…-ire
-a
-ite
kũkarĩ-
∅
raaaraza-
Imperative
Subjunctive
Indicative
-e
-e
-a
-e
-e
-
-e
-e/-(n)i
-
-e
-e
-
Imperfective
Perfective
-
-
-ga
-a
-a(ga)
-(y)e
hu-/si-
si-
ti-
si-
nge-/ngalihukajijapoki-
andinenaakya-
ngĩgaro-
nina-
15
12
17
12
Present
Present II
Pres. Continuous
Recent Past
Distant Past
Perfect
Past Perfect
Immediate Future
Near Future
Distant Future
Negation
Conditional
Habitual
Narrative
‘not yet’
‘even if’
‘if’
‘still’
optative
‘also’
The first section shows various tense morphemes: variants of past, present, and future.
For some languages (such as Gikuyu and Kinyarwanda) there are various past and future
tenses, depending on the temporal proximity to the speech event. For languages with
only one distinction for a particular tense, the form is listed in the tense closest to the
present. For example Swahili only has one past tense, which is listed in the “Recent Past”
row. The abbreviation ms for Luganda, indicates that a “modified stem” is used to indicate
the distant past, formed by a lexically-determined set of stem-changing operations. In
Gikuyu, the distant past is marked by the combination of a prefix and suffix, indicated
by a-…-ire. In Swahili, the present na- can also be used for present continuous.
14
1 Linguistic complexity: A case study from Swahili
The second category covers indicative, subjunctive, and imperative morphology, found
consistently among all of the languages. For Swahili, the final vowel -a is used as a general indicative mood marker.
The third category is aspect. Kinyarwanda and Gikuyu both have a distinction between perfective and imperfective, while Swahili and Luganda do not have morphology
for these aspectual distinctions.
All of the languages share cognate morphology for negation.
Other mood distinctions are covered in the final section of Table 6. This is reserved for
mood categories that are highly idiosyncratic meanings in particular languages, such as
morphology for meanings such as ‘not yet’ and ‘still’ in Swahili and Luganda, respectively. Another is the “optative” in Gikuyu, used for blessings and curses (Barlow 1960).
The narrative morpheme is used for verbs that are in a series during a narration of events.
There is no clear indication that any of these languages has a notably simpler system
of TAM morphology. Summing the number of morphological category distinctions made
in the four languages, it is clear that the inventory of distinctions is quite comparable for
all the languages, and Swahili is not noticeably less complex than any other language.
It is important to note the heterogeneity among the languages’ TAM morphology; few
morphemes are cognate, which makes it impossible to compare the phonological reduction among the languages, meaning that the phonological reduction of forms cannot be
measured for complexity.
6 Discussion: complexity and language contact
Data comparing the phonological inventory and morphological systems among Swahili,
Gikuyu, Kinyarwanda, Lingala, Haya, and, Luganda – as well as a comparison with ProtoBantu – show that there is no instance where clear decomplexification has occurred
in Swahili. In fact, in some instances, such as in consonant inventory, Swahili shows
more complexity that the other languages. In nearly all of the grammatical properties
discussed, Swahili is highly divergent from the other languages, with notable differences
in phonological inventory, such as a larger consonant inventory, a smaller vowel inventory, and irregularities with respect to stress and syllabification. Crucially, all phonological changes that have occurred have happened via natural sound changes, but at a faster
rate that than the other languages, i.e. Swahili is less similar to Proto-Bantu than the
other languages.
This grammatical situation fits neatly within recent studies of the typological and sociolinguistic literature on contact: language contact results in an increased rate of change,
and prolonged contact between two languages moves towards more linguistic complexity (Trudgill 2011). Prolonged contact with Arabic via the Omanis’ presence in Zanzibar
since the 13th century result in a strong change in the grammar of the language in comparison to other Bantu languages; however, it never blended with Arabic and became a
pidgin or creole. Mufwene (2001) and Mufwene (2003) also notes the divergent behavior
of Swahili when compared to other contact languages in Africa, showing that the exogamous use of Swahili has led to its adoption by the local population, which resulted in
15
Kyle Jerro
a relatively consistent use of Swahili. From this perspective, Swahili’s divergence from
the other languages is attributable to the specific contact situation of prolonged bilingualism with Arabic. Crucially, none of the comparison languages have engaged in this
kind of long-term bilingualism, accounting for grammatical differences between them
and Swahili.
In this paper, I have compared Standard Swahili as described in Ashton (1966) to the
standard varieties of several other varieties of East African Bantu languages. As just
noted, standard coastal Swahili has been in long-term contact with Arabic since the
13th century, and this contact resulting in expedited change (and, at times, complexification) of several grammatical features of the standard variety. Another prediction from
the literature on linguistic complexity is that simplification of grammar occurs when
adult learners attempt to learn a second language. Kusters (2003a) fleshes out this claim,
comparing Standard Swahili (the variety discussed in the present paper) to two other
varieties of Swahili that are used as lingua francas in areas where several adult speakers
of the languages speak it regularly, specifically, inland Kenyan Swahili and the Swahili
spoken in the trade town of Lubumbashi in the Katanga region of the Democratic Republic of the Congo. Crucially, both of these varieties have less prestige than the coastal
standard.
Kusters’ findings fit the typological pattern predicted: these two lingua franca languages show several reductions in category distinctions, morphophonological complexity and a reduction of inflectional information. For reasons of space, I refer the reader to
Kusters’ work, but the crucial point for the current discussion is that the three varieties of
Swahili are clear examples of the two kinds of second-language learning in contact areas.
Standard Swahili exemplifies the effects of long-term language contact, with acquisition
by young children: it has a radically divergent and at times more complex grammar than
related non-contact Bantu languages. Two other varieties of Swahili that have largely
been used as lingua francas by adult second-language speakers show systemic reduction in grammatical structure when compared with standard Swahili (Kusters 2003a).
Acknowledgements
I am indebted to Tony Woodbury, Pattie Epps, Peter Trudgill, and the anonymous reviews
for their invaluable comments during the development of this paper. I would also like to
thank the audience of the 45th Annual Conference on African Linguistics for their input
and suggestions. All errors remain the fault of the author.
Abbreviations
1
2
3
16
first person
second person
third person
asp
ben
pl
aspect
benefactive applicative
plural
sg
singular
1 Linguistic complexity: A case study from Swahili
References
Ashton, Ethel O. 1966. Swahili grammar. London: Longmans.
Baker, Mark C. 2003. Linguistic differences and language design. Trends in Cognitive
Science 7. 349–353.
Bakker, Peter. 1997. A language of our own: The genesis of Michif, a mixed Cree-French
language of the Candadian Métis. Oxford: Oxford University Press.
Barlow, A. Ruffel. 1960. Studies in Kikuyu grammar and idiom. Edinburgh: William Blackwood & Sons, Ltd.
Bleek, Wilhelm. 1869. A comparative grammar of South African languages. London: Trübner.
Byarushengo, Ernest Rugwa, Alessandro Duranti & Larry M. Hyman. 1977. Haya grammatical structure. Los Angeles: University of Southern California Press.
Comrie, Bernard. 2008. Inflectional morphology and language contact, with special reference to mixed languages. In Peter Siemund & Noemi Kintana (eds.), Language contact
and contact languages, 15–32. Amsterdam: John Benjamins.
Dahl, Östen. 2004. The growth and maintenance of linguistic complexity. Amsterdam: John
Benjamins.
Engstrand, Olle & Abdulaziz Y Lodhi. 1985. On aspiration in Swahili: Hypotheses, field
observations, and an instrumental analysis. Phonetica 42. 175–185.
Givón, Talmy. 1984. Syntax: A functional-typological introduction. Amsterdam: John Benjamins.
Givón, Talmy & Masayoshi Shibatani (eds.). 2009. Syntactic complexity: Diachrony, acquisition, neuro-cognition, evolution. Amsterdam: John Benjamins.
Guthrie, Malcom. 1966. Grammaire et dictionnaire de Lingala. Westmead, England: Gregg
Press Ltd.
Guthrie, Malcom & John F Carrington. 1988. Lingala grammar and dictionary: English–
Lingala, Lingala–English. London: Baptist Missionary Society.
Halpert, Claire. 2012. Argument licensing and agreement in Zulu. MIT dissertation.
Hay, Jennifer & Laurie Bauer. 2007. Phoneme inventory size and population size. Language 83. 388–400.
Heine, Bernd & Tania Kuteva. 2005. Language contact and grammatical change. Cambridge: Cambridge University Press.
Hyman, Larry M. 2003. Segmental phonology. In Derek Nurse & Gerard Philippson (eds.),
The Bantu languages, 42–58. New York: Routledge.
Hyman, Larry M. & Francis X. Katamba. 1991. Augment in Luganda tonology. Journal of
African Languages and Linguistics 12. 1–46.
Hyman, Larry M. & Francis X. Katamba. 1993. The augment in Luganda: Syntax or pragmatics? In Sam Mchombo (ed.), Theoretical aspects of Bantu grammar 1, 209–256. Stanford: CSLI.
Jerro, Kyle. 2015. Revisiting object symmetry in Bantu. In The Selected Proceedings of the
44th Annual Conference on African Linguistics, 130–145. Somerville, MA: Cascadilla
Proceedings Project.
17
Kyle Jerro
Katamba, Francis X. 2003. Bantu nominal morphology. In Derek Nurse & Gerard Phillippson (eds.), The Bantu languages, 103–120. London: Routledge.
Kimenyi, Alexandre. 1980. A relational grammar of Kinyarwanda. Los Angeles: University of California Press.
Kirwan, Brian Edmond Renshaw & Peter A. Gore. 1951. Elementary Luganda. Kampala:
Uganda Bookshop.
Kusters, Wouter. 2003a. Linguistic Complexity: The Influence of Social Change on Verbal
Inflection dissertation.
Kusters, Wouter. 2003b. The fate of complex languages: Classical Arabic and Old Norse
in the age of globalisation. Nordlyd 31. 275–289.
Maddieson, Ian. 2003. The sounds of the Bantu languages. In Derek Nurse & Gerard
Philippson (eds.), The Bantu languages, 15–41. New York: Routledge.
McWhorter, John. 2008. Why does a language undress? Strange cases in Indonesia. In
Matti Miestamo, Kaius Sinnemäki & Fred Karlsson (eds.), Language complexity: Typology, contact, change, 167–190. Amsterdam: John Benjamins.
McWhorter, John. 2011. Linguistic simplicity and complexity: Why do languages undress?
Berlin: Mouton de Gruyter.
Meeussen, Achille Emile. 1967. Bantu grammatical reconstructions. Africana Linguistica
3. 80–122.
Miestamo, Matti. 2008. Grammatical complexity in a cross-linguistic perspective. In
Matti Miestamo, Kaius Sinnemäki & Fred Karlsson (eds.), Language complexity: Typology, contact, change, 23–42. Amsterdam: John Benjamins.
Miestamo, Matti, Kaius Sinnemäki & Fred Karlsson (eds.). 2008. Language complexity:
Typology, contact, change. Amsterdam: John Benjamins.
Mufwene, Salikoko S. 2001. The ecology of language evolution. Cambridge: Cambridge
University Press.
Mufwene, Salikoko S. 2003. Contact languages in the Bantu area. In Derek Nurse & Gerard Philippson (eds.), The Bantu languages, 195–208. New York: Routledge.
Myers, Scott & Megan Crowhurst. 2006. Phonology: Case studies. http://www.laits.utexas.
edu/phonology/.
Nichols, Johanna. 2009. Linguistic complexity: A comprehensive definition and survey.
In Geoffrey Sampson, David Gil & Peter Trudgill (eds.), Language complexity as an
evolving variable, 110–125. Oxford: Oxford University Press.
Nurse, Derek. 2003. Aspect and tense in Bantu languages. In Derek Nurse & Gerard
Philippson (eds.), The Bantu languages, 90–102. New York: Routledge.
Pinker, Steven. 1994. The language instinct. New York: W. Morrow.
Pinker, Steven & Paul Bloom. 1990. Natural language and natural selection. Behavioral
and Brain Sciences 13. 707–726.
Russell, Joan. 2003. Teach yourself Swahili. Chicago: McGraw-Hill Companies, Incorporated.
Sampson, Geoffrey. 2009. A linguistic axiom challenged. In Geoffrey Sampson, David
Gil & Peter Trudgill (eds.), Language complexity as an evolving variable, 1–18. Oxford:
Oxford University Press.
18
1 Linguistic complexity: A case study from Swahili
Schadeberg, Thilo C. 2003a. Derivation. In Derek Nurse & Gerard Philippson (eds.), The
Bantu languages, 71–89. New York: Routledge.
Schadeberg, Thilo C. 2003b. Historical linguistics. In Derek Nurse & Gerard Philippson
(eds.), The Bantu languages, 143–163. New York: Routledge.
Trudgill, Peter. 2004. Linguistic and social typology: The Austronesian migrations and
phoneme inventories. Linguistic Typology 8. 305–320.
Trudgill, Peter. 2009. Sociolinguistic typology and complexification. In Geoffrey Sampson, David Gil & Peter Trudgill (eds.), Language complexity as and evolving variable,
98–109. Oxford: Oxford University Press.
Trudgill, Peter. 2011. Sociolinguistic typology: Social determinants of linguistic complexity.
Oxford: Oxford University Press.
Zerbian, Sabine & Manfred Krifka. 2008. Quantification across Bantu languages. In
Lisa Matthewson (ed.), Quantification: A cross-linguistic perspective, 383–415. Bingley:
Emerald Publishing Group.
19
Chapter 2
Inter-party insults in political discourse
in Ghana: A critical discourse analysis
Emmanuel Amo Ofori
University of Florida/University of Cape Coast
In recent times, politics in Ghana has become the politics of personal attack, vilification,
and insults. Various attempts have been made to stop this brand of politics, including one
spearheaded by the Media Foundation for West Africa, which releases a weekly report to
the general public aimed at shaming politicians who are involved in the politics of insults. If
a country could go to the extent of shaming politicians involved in politics of insults, then it
shows how the issue of intemperate language has become entrenched in Ghanaian political
discourse. Thus, there is a need to conduct a thorough analysis of the realization of insults
in Ghanaian political discussion. Utilizing a Critical Discourse Analysis approach, this paper analyzes the underlying ideologies in the representation of insults in pro-New Patriotic
Party (NPP) and National Democratic Congress (NDC) newspapers. It further compares and
contrasts the use of insults in the newspapers.
1 Introduction
Ghana is a democratic country, and this has earned the West African country enviable
recognition in the world. Ghana experienced her stable democratic dispensation in 1992
after a series of military take-overs/coup d’états from the period of independence until
1992 (4th republic). The democratic practices in Ghana are still in the infant stages and
therefore it could be considered as an emerging democratic state. Since 1993, political discussions in Ghana have centered on various topics, such as the economy, social policies,
employment, youth development, education, national security, and health. However, in
recent years, Ghanaian political discourse has become a discourse of personal attack, vilification, and insults. Many have argued that the surge of insults in Ghanaian political
discussions is due to the liberalization of the media in Ghana (Owusu 2012; Marfo 2014).
This stems from the fact that before 1992, Ghana did not have many radio and television
stations, newspapers and online websites. Currently, there are numerous radio stations
and newspapers in Ghana, and most of the insults emanate from politicians through
the various media outlets. These outlets have their own interest in promoting certain
Emmanuel Amo Ofori. Inter-party insults in political discourse in Ghana: A critical discourse
analysis. In Jason Kandybowicz, Travis Major, Harold Torrence & Philip T. Duncan (eds.), African
linguistics on the prairie: Selected papers from the 45th Annual Conference on African Linguistics,
21–35. Berlin: Language Science Press. DOI:10.5281/zenodo.1251710
Emmanuel Amo Ofori
ideologies and political positions. This is seen in how the media present their audience
with “a steady supply of problems and crisis, and it may be their interest to exaggerate
a problem, fostering the impression that there is a crisis and not just business as usual”
(Cameron 2012: 83). It may seem that they are alerting the public to the surge of intemperate language; however, it is a subtle way of promoting an ideology or political position.
Therefore the representation of insults from opponents is publicized or foregrounded
not to alert the public about the problem of insults, but to put a political spin on it. The
us versus them dichotomy was seen in media reportage of insults in pro-NPP and NDC
newspapers. Thus, van Dijk’s (1998) concept of ideological square, which is expressed in
terms of emphasizing the positive actions of what a media institution considers the ingroup and deemphasizing its negative actions, while, on the other hand, deemphasizing
the positive actions of the outgroup, and emphasizing its negative actions, is applied in
the analysis of the use of insults in Ghanaian political discourse.
In Ghana, politicians own some of the radio stations and newspapers as the means
to disseminate the ideology and philosophy of their respective political parties. This is
an attempt on the part of the political parties to control the media. Different groups
compete in order to control the media as an instrument of social power, or an Ideological State Apparatus (ISA) in the sense of Althusser (1971), to legitimate and naturalize
their ideologies, beliefs, and values (van Dijk 1995). Anyone who controls the media, to
some extent, controls the minds of its listeners. This is because the media is seen as a
major source of information. Therefore, this paper analyzes the underlying ideologies
in the representation of insults in pro-New Patriotic Party (NPP) and National Democratic Congress (NDC) newspapers. It further compares and contrasts the use of insults
in newspapers.
1.1 NPP and NDC
Ninsin (2006) observed that when the ban on political parties was lifted in May 1992, by
November the same year 13 political parties had been formed and registered. However,
the two dominant parties that have survived since 1992 are the National Democratic
Congress (NDC) and the New Patriotic Party (NPP).
The NPP, the current party in opposition, emerged from an old political tradition dating back to the United Gold Coast Convention (UGCC) and United Party (UP) of the
Danquah-Busia tradition. They fought for independence with Dr. Kwame Nkrumah’s
Conventions Peoples Party (CPP). The UGCC and the UP metamorphosed into the NPP
in 1992 when the country returned to civilian rule after 11 years of military rule, in order
to contest the 1992 December elections. Their political ideology is founded on capitalism,
and they believe in privitalization, rule of law, and democracy. In short, they see the private sector as an engine of growth. The NPP lost both the presidential and parliamentary
elections in 1992 and 1996. They won the 2000 and 2004 elections, and lost to the NDC
in 2008 and 2012 in one of the most closely contested presidential elections in Ghana’s
history. The NPP lost by a margin of 0.46% in 2008.
22
2 Inter-party insults in political discourse in Ghana: A critical discourse analysis
The NDC, the current government in power, is one of the newest parties in Ghanaian
politics. It was formed in 1992 from the Provincial National Defence Council (PNDC)
military regime with Flt. Lt. Jerry John Rawlings as its leader. The PNDC overthrew
a constitutionally elected government, the People’s National Party (PNP), which ruled
Ghana from 1979 to 31st December 1981, and ruled Ghana from 1981 to 1992. During the
return to civilian rule, the PNDC metamorphosed into the NDC. The NDC won both the
1992 and 1996 elections. They then lost to the NPP in 2000 and 2004, and won the 2008
and 2012 elections. The political ideology of the NDC is founded on social democracy.
1.2 Data collection
The data for this study were obtained from reports in pro-NPP and NDC newspapers. The
newspapers are Daily Guide, The Daily Searchlight, The New Statesman, and The Chronicle
(pro-NPP newspapers);The Informer, The Democrat, The Palaver, The Al-Hajji, The National
Democrat, The Catalyst, The New Voice, Daily Post, Daily Heritage, Radio Gold online (proNDC newspapers). In all, a total of 78 news articles were gathered from 2012 to February
2014. The articles were sampled and analyzed, using Fairclough’s discourse-as-text (1989,
1992a, 1995a,b, 2000, 2003) and van Dijk’s (1998) concept of ideological square.
1.3 Definition of insult
There are various definitions of insult. According to Aristotle, “insult is belittlement. For
an insult consists of doing or saying such things as involve shame for the victim, not for
some advantage to oneself other than these have been done but for the fun of it” (Aristotle
Rhetoric cited in Yiannis 1998). Aristotle’s definition of insult focuses on shame, for the
fun of it, and it is a form of belittling the target.
In this paper, the working definition of insult adopted is a modification of Yiannis’s
(1998: 3) social psychology definition of insults, which considers insult as “a behavior or
discourse, oral or written, which is perceived, experienced, constructed and at all times
intended as slighting, humiliating, or offensive. Insult can also be verbal, consisting of
mocking invective, cutting remarks, negative stereotypes rudeness or straight swearing”.
I therefore define insult as “a behavior or discourse, oral or written, direct or indirect,
gestural or non-gestural, which is perceived, experienced and most of the time intended
as slighting, humiliating, or offensive, which has the potential of psychologically affecting not only the addressee or target but also his/her associates.” This definition should
not be taken as universal because there is no universal measure of insults. The yardstick
to measure insults differs from society to society and also from culture to culture.
2 Critical Discourse Analysis (CDA)
The revolving idea of CDA is power, and it analyzes opaque as well as transparent structural relationships of dominance, discrimination, power and control as manifested in
language (Wodak 2001: 2). van Dijk (2001: 96) also postulates that CDA focuses on social
23
Emmanuel Amo Ofori
problems, especially on the role of discourse in the production and reproduction of power
abuse or domination. This means that it focuses not only on linguistic elements per se,
but also complex social phenomena that have semiotic dimensions (Wodak & Meyer
2009). In effect, the overall aim of CDA is linking linguistic analysis to social analysis
(Wood & Kroger 2000: 206). CDA aims at making visible and transparent the instrument
of power, which is of increasing importance in the contemporary world. It is very critical of the relationship between language, discourse, speech, and social structure. As the
dimensions of CDA include “the object of moral and political evaluation, analyzing them
should have effects on society by empowering the powerless, giving voices to the voiceless, exposing power abuse, and mobilizing people to remedy social wrongs” (Blommaert
2005: 25). These are the main concerns in analyzing insults in Ghanaian public political
discourse: Who has access to the media? Who controls the media? What are the ideological standpoints of the media in Ghana? Whose agenda are they propagating? These are
some of the questions that CDA tries to uncover in contemporary societies that relate
directly to the present study.
The media discourse in Ghana has changed drastically in that before 2001 it was very
difficult for media personnel to operate. This was due to the various laws governing
media practices in Ghana. Even the ones that existed were so polarized that they were
divided into two distinct genres: state press and private press (Hasty 2005). The state
press were praise singers of the government. They published stories that projected the
development, inspirational rhetoric and policies of the government. The private press
was sometimes the opposite of the state press. They revealed the profligate spending,
abuse of power, and social inequality attributed to the government. In analyzing the
underlying ideologies in the representation of insults in pro-NPP and NDC newspapers,
these developments must be taken into consideration. The two CDA approaches applied
in this paper are Fairclough’s discourse-as-text and van Dijk’s ideological square.
2.1 Fairclough’s discourse-as-text
Fairclough situates his theory of social-discoursal approach in Halliday’s Systemic Functional Linguistics (SFL) and also draws on critical social theories, such as Foucault’s concept of order of discourse, Gramsci’S concept of hegemony, Habermas’ concept of colonization of discourses and many others. To fully understand the interconnectedness between languages, social and political thought, Fairclough proposes a three-dimensional
approach to analyzing discourse. These are: discourse-as-text, discourse-as-discursivepractice and discourse-as-social practice. Since the analysis in this paper is based on
discourse-as-text, I elaborate on it below (for detailed discussions on the other two dimensions see Fairclough 1989, 1995).
Discourse-as-text involves the analysis of the way propositions are structured and the
way they are combined and sequenced (Fairclough 1995). Here, the analyst examines the
text in terms of what is present and what could have been present but is not. The text, and
some aspects of it, is the result of choice, that is, the choice to describe a person, an action
or a process over another; the choice to use one way of constructing a sentence over an
24
2 Inter-party insults in political discourse in Ghana: A critical discourse analysis
alternative; the choice to include a particular fact or argument over another. According
to Fairclough (1995: 57), choice in text “… covers traditional forms of linguistic analysisanalysis of vocabulary and semantics, the grammar of sentences and smaller units, and
the sound system (phonology) and writing system. But it also includes analysis of textual
organization above the sentence, including the ways in which sentences are connected
(cohesion) and aspects like the organization of turn-taking in interviews or the overall
structure of a newspaper article”. The application of textual analysis in CDA does not
mean just focusing on the linguistic form and content; rather, it is the function that such
elements play in their use in the text. Hence, the traditional forms of linguistic analysis
should be analyzed in relation to their direct or indirect involvement in reproducing or
resisting the systems of ideology and social power (Richardson 2007).
2.2 Van Dijk’s concept of ideological square
One prominent feature of van Dijk’s socio-cognitive approach is the concept of ideological square. It is about how different social groups project themselves positively and
represent others negatively. The structures of ideologies are represented along the lines
of an us verses them dichotomy, in which members of one social group present themselves in positive terms, and others in negative terms. There is polarization of how media institutions emphasize the positive actions of ingroup members and deemphasize
its negative action on one hand, and deemphasize the positive action of the outgroup
while emphasizing its negative actions. The ideological square consists of four moves: (1)
express/emphasize information that is ‘positive’ about us; (2) express/emphasize information that is ‘negative’ about them; (3) suppress/deemphasize the information that is
‘positive’ about them; and (4) suppress/deemphasize information that is ‘negative’ about
us. Any property of discourse that expresses, establishes, confirms or emphasizes a self
interested group opinion, perspective or position, especially in a broader socio-political
context of social struggle, is a candidate for special attention in ideological analysis (van
Dijk 1998).
2.3 Ideology
Ideology is defined as systems of ideas, beliefs, practices, and representations, which
work in the interest of a social class or cultural group. Gramsci sees ideology as “tied to
action, and ideologies are judged in terms of their social effects rather than their truthvalues” (Fairclough 1995: 76). Ideology has the potential to become a way of creating and
maintaining unequal power relations, which is of central concern to discourse analysts
who take a “particular interest in the ways in which language mediates ideology in a
variety of social institutions” (Wodak 2001: 10). van Dijk (1998: 3) also defines ideology
as political or social systems of ideas, values or prescriptions of a group that have the
function of organizing or legitimating the actions of the group. The use of language
reflects a person’s philosophical, cultural, religious, social, and political ideology. Hence,
ideology affects the way one talks, argues, and reacts.
25
Emmanuel Amo Ofori
3 Textual analysis
Textual analysis is Fairclough’s first dimension of his three dimensional framework. The
linguistic tools employed under this dimension in the analysis are lexicalization and
predication, presupposition and verbal process.
3.1 Lexicalization and predication
Lexicalization involves the choice or selection and the meanings of words used to refer
to social actors. A typical lexical analysis looks at the denotation (the literal or primary
meaning of words) and connotation (the various senses that a word invokes in addition to
its literal or primary meaning). This type of analysis is important because “words convey
the imprint of society and of value judgments in particular” (Richardson 2007: 47). There
is a strong relationship between lexicalization and ideology, as in the use of expressions
such as ‘terrorist’ versus ‘freedom fighter’ for example. This suggests that language users
have several choices of words to refer to the same persons, groups, social relations or
issues, and in most or all cases these carry heavy semantic and ideological loads. The
words in a text that communicate messages about subjects or themes in newspapers are
framed ideologically. Thus “vocabulary encodes ideology, systems of beliefs about the
way the world is organized” (Fowler 1987: 69).
The paper also analyzes predication. Wodak & Meyer (2009: 27) define predication
strategies as terms or phrases that “appear in stereotypical, evaluative attribution of
positive or negative traits and implicit or explicit predicates”. Reisigl & Wodak (2001: 54)
also see predicational strategies as “the very basic process and result of linguistically
assigning qualities to persons, animals, objects, events, actions and social phenomena”.
Predicational strategies are not used arbitrarily: there are hidden ideologies in the various forms or phases. It also reveals the ‘Us’ versus ‘Them’ dichotomy that shows positive
predications for the ingroup and negative predications for the outgroup. Indeed, “predication is used to criticize, undermine and vilify certain social actors, sometimes with
potential dangerous consequences” (Richardson 2007: 53).
NPP newspapers reported insults with lexicalization and predication from people they
considered as ingroup members targeted at the outgroup members. These included:
(1)
Specialist in lies and propaganda shouldn’t be managing our economy Bawumia Daily searchlight, October 2, 2012
(2)
J. J. tainted with blood - Daily Guide, December 5, 2013
(3)
Greedy bastards and babies with sharp teeth - Rawlings The New Statesman,
April 3, 2014
(4)
Thieving Mahama caught - Daily Searchlight, October 2, 2012
The insult in (1) is a quote from Dr. Bawumia, a running mate to the presidential candidate of the main opposition party in Ghana. He has a PhD in economics and believes
26
2 Inter-party insults in political discourse in Ghana: A critical discourse analysis
that the NPP has the experts to manage the economy. However, he believes that those
managing the economy do not have the requisite expertise to manage it; they rely only
on lies and propaganda, meaning they are not being truthful with the true state of the
economy. Looking at the background of Dr. Bawumia, it would be difficult for the public
to scrutinize his comment and it may not be possible for readers to understand that this
is an attempt to convince the electorate that if the NPP is voted into power they will
make the economy better. The underlying idea behind the report of this insult is that
the ingroup has the experts to manage the economy better than the outgroup. Therefore,
the newspaper reports the insults from the ingroup member seeking to negatively portray the outgroup as people who do not have the technical know-how to manage the
economy.
Example (2) is a predication that projects the founder of the NDC (an outgroup member) as a killer. J. J. Rawlings has been accused several times of being responsible for the
killing of the three Heads of State and three Supreme Court judges in Ghana. Therefore,
the report of this insult from the NPP newspaper is a way to remind the public that the
founder of the NDC is a known killer. Thus, the underlying idea behind the report of this
insult is to negatively portray the outgroup member as a killer and not worth listening
to.
The predicational insult in (3), greedy bastards and babies with sharp teeth was a comment made by the former President Rawlings to his own party members vilifying him.
Expanding on this insult, it can be analyzed in two ways: first, it is used to describe the
activities of the NDC with regards to corruption. Secondly, the second part of the insult,
babies with sharp teeth, is a metaphor that describes the behavior of the young ones in
the NDC. Rawlings made this comment at a time when most of the youth in the NDC
were insulting him. So, his metaphor babies with sharp teeth refers to these youth who
have the penchant toward behavior of vilifying and insulting people. Within the Ghanaian culture, kids or babies are not supposed to engage in adult communication, let alone
to insult adults. This metaphor portrays the young people within the NDC as having
outgrown their wings and vilifying the adults in the party. Therefore, pro-NPP newspapers reported this insult among others to expose the confusion within the camp of the
outgroup, and to show that they are not the only people saying the outgroup members
are corrupt; the founder of their party and members of their party concur with them.
In (4), the Daily Searchlight paper describes the president of Ghana, John Mahama, as
a thief, for stealing the 2012 presidential election, which the NPP contested in court and
lost. Calling the president a thief was to bring to the attention of the NPP supporters
that the party won the 2012 elections, and that their loss was a typical case of a stolen
verdict. Therefore, the insult was aimed at satisfying the aggrieved ingroup members
and portraying the outgroup as thieves.
Similarly, pro-NDC newspapers used lexicalization and predicational strategy to report insults from the ingroup members targeted at the outgroup members.
(5) No ‘patapaa’ President Mahama warns losers in December election - Ghana
Palaver, September 14–16, 2012
27
Emmanuel Amo Ofori
(6) Arrogant Kan Dapaah running a one man show - Ghana Palaver, July 20, 2012
(7)
Akuffo-Addo is wicked and not worth dying for - Ghana Palaver, June 20, 2012
(8)
“T𝜀 Ni” can’t govern us. - The Al-Hajj, August 16, 2012
The lexical item patapaa in example (5) was uttered by the current president of Ghana,
John Mahama, to advise all losers of the 2012 election, and this was reported by the Ghana
Palaver newspaper. Patapaa is an Akan word, which means “a person who is violent or
a violent behavior”. It was used to describe the losers in the December elections. Though
the word was part of a comment made to advise all losers, knowing that Ghana’s election
is a contest between the NDC and NPP, this insult was directed at the doorsteps of the
main opposition party and most importantly to the 2012 presidential candidate of the
NPP, Nana Akuffo-Addo. Because he was alleged to have said prior to the 2012 elections
that the elections would be “all die be die”, he was criticized for warmongering. Thus,
the report of the president’s insult is to negatively present the outgroup member, the
NPP presidential candidate, as a violent person who would not accept the 2012 election
results and would plunge the country into chaos if he lost; by extension, it is implied he
is not even qualified to be a president.
In (6), the Ghana Palaver newspaper reports insults from NDC members of the Public Accounts Committee (PAC) of Parliament that were tasked to investigate financial
malfeasance in the public service. Parliamentary regulations provide the opportunity
for an MP from the opposition party to chair the (PAC) so that there is a fair investigation of government officials. The NDC members accused the chairman of the committee,
Albert Kan Dapaah, an outgroup member, of being arrogant and running a one-man show
because he threatened to cause the arrest of an ingroup member, Alfred Agbesi Woyome.
The newspaper reports this predicational insult from the NDC members of the committee to negatively portray the outgroup member as someone who does not consult them
before making decisions, and has therefore taken an arrogant posture. The publication
of the insult is clearly meant to discredit the outgroup member.
The Ghana Palaver’s predicational insult in (7) is targeted at Nana Akuffo-Addo, the
NPP presidential candidate for 2008 and 2012 elections when he continued with his campaign at a time when the late former Vice President, Aliu Mahama, was admitted to the
hospital. Akuffo-Addo was accused of not showing enough compassion and abandoning the former Vice President to die. The ingroup candidate is reported to have asked
for prayers for the late vice president from Ghanaians. Therefore, the paper described
Akuffo-Addo negatively as wicked and not qualified to be president of Ghana. The ingroup candidate is preferred over the outgroup candidate, revealing a group polarization
between NPP and NDC newspapers.
The Al-Hajj newspaper reported the insult in (8) from a taxi driver who claimed to
have heard some Akans using the lexical item “t𝜀 ni” to insult the president. This word
needs elaboration. The correct spelling of the word is “tani,” an Akan term used to insult
people from the Northern part of Ghana. Recall that the NDC presidential candidate,
John Mahama is a Northerner, and the NPP is perceived as an Akan dominated party.
28
2 Inter-party insults in political discourse in Ghana: A critical discourse analysis
So, for this term to surface on the front page of an NDC paper is a way of turning the
people of the North against the NPP. The paper portrays to readers that the outgroup is
presenting the ingroup presidential candidate as someone who is ethnically unfit to lead
the country. This portrays the outgroup negatively for using tribal and ethnic sentiments
against the ingroup candidate, while the ingroup is implicitly presented positively for not
whipping up ethnic sentiments.
3.2 Presupposition
Presupposition is a “taken-for-granted, implicit claim embedded within the explicit meaning of a text or utterance” (Richardson 2007: 63). Wodak (2007: 214) provides a broader
picture of presupposition: “presupposed content is, under ordinary circumstances, unless there is a cautious interpretive attitude on the part of the hearer, accepted without
(much) critical attention (whereas the asserted content and evident implicatures are normally subject to some level of evaluation)”. The claims are not critically evaluated and are
generally considered to be true regardless of whether the sentence is true. It is a useful
strategy in political discourse because it makes it difficult for the audience to identify or
reject views communicated in this way. That is, it persuades people to take for granted
something which is actually open to debate (Bayram 2010).
Pro-NPP and NDC newspapers utilized presupposition strategy using the authorial
voice and non-politicians to describe people they considered to be outgroup members.
Below are examples from pro-NPP papers:
(9) In his usual propaganda style - The Chronicle, July 19, 2012
(10) Send people who can make intellectual debate daily - Daily Guide, December 12,
2013
(11) Former President Jerry John Rawlings has suddenly found his voice - Daily
Guide, December 10, 2010
The Chronicle paper used the presupposition in (9) to report an insult hurled on one
of their reporters by Deputy Minister for Information, James Agyenim Boateng when he
asked a question on whether the Constitutional Review Committee (CRC) was considering putting a clause in the constitution about health status of presidents and presidential
candidates. Recall that it was during this time that the late president Mills was alleged
to have been taken ill. The paper therefore presented members of the outgroup as people fond of using propaganda. Employing the lexical item usual by The Chronicle paper
indicates that the outgroup member is noted for his habitual propaganda style.
In (10), the presupposition was reported from a source that presented the outgroup
negatively. Fiifi Banson, a broadcast journalist with Peace FM, a radio station in Accra, is
reported by the Daily Guide paper to have uttered that presupposition. He is presented
as “an award winning Ghanaian broadcaster” to portray the genuineness of the presupposition. The meaning behind the presupposition is that the crop of panelists sent by the
NDC to Peace FM are not astute enough to present the agenda of the party.
29
Emmanuel Amo Ofori
The former President Rawlings is also presented in the Daily Guide paper in (11) as
suddenly finding his voice when he insulted former president Kufuor as an “autocratic
thief”. The use of sudden presupposes that Rawlings has been quiet for sometime and
was now speaking. His sudden voice, however, was directed at an ingroup member not
the outgroup, which the paper believes is noted for corruption. Therefore, the use of this
presupposition is to negatively present the former president, an outgroup member, as directing his attention towards the wrong person. In sum, pro-NPP newspapers presented
the outgroup members negatively using presupposition from both the authorial voice
(depicting the ideology of the papers) and other sources.
Pro-NDC newspapers also employed presupposition strategy to present the outgroup
negatively. They used presuppositions from the authorial voice and members in the ingroup. Some of the presuppositions included:
(12)
Loose-talking ‘Genocide’ MP on the loose again - The Catalyst, September 7, 2012
(13)
NPP turned Ghana into a cocaine country - Felix Kwakye-Ofosu country
radiogold.com, June 7, 2013
(14)
True NPP old evil Dwarfs at work - Ghana Palaver, July 20, 2012
The Catalyst reported this presupposition in (12), using the authorial voice, to represent Kennedy Agyapong, the outgroup MP, negatively for insulting the President, John
Mahama, the entire membership of NDC and the police service. The use of the lexical
item again in the report presupposes that the MP’s loose talking is not the first time. That
is, the MP is known for his habitual loose talk. He is even described as a genocide MP,
a term which negatively portrays the MP. Thus, the presupposition and the description
are used to negatively present the outgroup member.
Radio Gold reported the presupposition in (13) from a deputy minster for information,
Felix Ofosu Kwakye who described as unfortunate attempts by the NPP to link the arrest
of the Managing Director of SOHIN Security in the United States for drug trafficking to
the Mahama administration. The word turned presupposes that before the NPP came to
power there was nothing like cocaine in Ghana; therefore, the NPP are responsible for
turning Ghana into a cocaine country. The ingroup’s administration is implicitly presented positively for not being responsible for the cocaine business in the country.
The Ghana Palaver paper employed the presupposition in (14) to insult the outgroup.
The former President Rawlings was the first person to use the description old evil dwarfs
to refer to some members of the NDC. For an NDC newspaper to use this same description with the adjective true to refer to the outgroup is interesting. The adjective true is
used to qualify the noun phrase NPP old evil dwarfs. Richardson (2007) indicates that
the use of nouns and adjectives to modify noun phrases trigger presuppositions he calls
nominal presuppositions. So, true NPP old evil dwarfs is a nominal presupposition meant
to present the outgroup negatively. True presupposes that the NPP are indeed the real
or actual old evil dwarfs, not the NDC. Therefore, the nominal presupposition was used
to present the outgroup negatively. In sum, pro-NDC newspapers employed presuppositions, using authorial voice and the ingroup to present the outgroup negatively.
30
2 Inter-party insults in political discourse in Ghana: A critical discourse analysis
3.3 Verbal process
Verbal process is “any kind of symbolic exchange of meaning” as well as predicates of
communication (Halliday 1985). This means that they represent the action of talking,
saying and communicating. Journalists use verbal processes to introduce the speech of
people they are reporting on, and this can reveal the feelings and the attitudes of the
journalists towards the people they consider important whose words or actions they
report. Such reportage can be used to marginalize others and focus readers’ attention
towards the direction of the reporter. Thus, “choosing certain verbal process rather than
others, the producer of a text is able to foreground certain meanings in discourse while
others are suppressed” (Chen 2005: 34).
Chen (2005), following Halliday’s analysis of verbal process, proposed three sub-categories of verbal process. The first is negative verbal process, which demonstrates a certain negative feeling on the part of the writer towards the person whose words the verbal
process is used to introduce. Examples of such verbs are ‘insisted’, ‘denied’, ‘claimed’, ‘admitted’, ‘complained’. The second is positive verbal process, which is used to promote
in a reader the feeling that the person whose words are being reported is wise, authoritative, benign or in some other sense positive. Examples are ‘pointed out’, ‘announced’,
‘explained’, ‘declared’, ‘indicated’, and ‘urged’. The last is neutral verbal process, where
the writer’s choice of verb does not indicate an endorsement or disparagement of what
the person being reported is saying. Examples include ‘said’, ‘told’, ‘described’, ‘asked’,
‘commented’.
Pro-NPP and NDC newspapers employed different verbal processes to report insults
from people they considered ingroup positively and those they considered outgroup
negatively. The analysis of the data showed that pro-NPP newspapers employed negative
and neutral verbs to report insults directed at the outgroup. However, NDC newspapers
adopted positive, negative and neutral verbs to report insults directed at the outgroup.
Table 1 sums up all the verbal processes employed by both NPP and NDC Newspapers.
Table 1: NPP and NDC verbal process.
newspapers
positive
NPP
NDC
explain
Confirm
negative
neutral
lashed out
Jabbed
blasted
accused
describe
challenge
say
warn
blast
lashed-out
condemned
describe
say
In sum, positive verbal processes were not very common; they only appeared in the reports of pro-NDC newspapers, showing that positive ingroup representation was more
31
Emmanuel Amo Ofori
common in pro-NDC newspapers compared to NPP newspapers. However, negative
other-representation manifested in both NPP and NDC newspapers.
4 Discussion
In this section, I expatiate on the textual analysis to investigate the broader sociopolitical and sociocultural context. Drawing on van Dijk’s (1995) ideological square, I discuss
the various ideological structures utilized by both pro-NPP and NDC newspapers on the
textual analysis to represent the ingroup positively and the outgroup negatively (group
polarization), paying particular attention to the sociopolitical context that necessitated
this polarization. The two ideological structures employed by pro-NPP and NDC newspapers are negative lexicalization and predications and detailed description. These revealed
group polarization between the two dominant political parties in Ghana.
In the first place, the lexical forms that are used to describe the political opponents
show Us verses Them dichotomy. Pro-NPP newspapers for example employed the lexical
item “propaganda” from Dr. Bawumia to refer to the managers of the Ghanaian economy.
According to the United States Institute for Propaganda Analysis (USIPA) (2001), the
word propaganda “is an expression of opinion or action by an individuals or groups
deliberately designed to influence opinions or actions of individuals with reference to
predetermined ends”. They further state that the main idea of a propagandist is to put
something across, either good or bad. The USIPA provides a list of seven propaganda
devises. These are: (1) name calling, (2) glittering generalities, (3) transfer, (4) testimonial,
(5) plain folks, (6) card stacking, and (7) band wagon (for detailed discussion of these
devises, see USPIA 2001). The pro-NPP newspapers, therefore, published this lexical item
for ideological purposes, that is, to present the outgroup as: (1) not giving the true state
of the economy and (2) bad handlers of the economy. However, they portrayed to the
reader that the NPP would be transparent and have the needed expertise to manage the
economy better than the outgroup. This reveals group polarization, in that, the ingroup
is presented positively as transparent and good managers of the economy while the outgroup is presented negatively as liars and bad handlers of the economy.
Pro-NDC newspapers also employed lexical items “patapaa” and “tani”. The current
president of Ghana, John Mahama prior to the 2012 general elections, used the word
“patapaa”, which means “a violent person or violent behavior or a thug”. The meaning
can be extended to “someone who uses force to claim what does not belong to him/her”.
This word was uttered to advise all losers of 2012 elections. However, this was an insult
directed at the presidential candidate of the NPP, describing him as violent, adopting a
“patapaa” stance to win the elections. NDC newspapers reported this negative lexicalization from the president to present the presidential candidate of the outgroup as violent
while maintaining that the ingroup candidate is peaceful and not violent.
NDC newspapers employed the term “tani”, an Akan word that is normally uttered to
insult those from the Northern part of Ghana. There is no consensus as to the meaning
of this word. While Agyekum (2010) argues that it is a derogatory term for people who
move in pairs (ntafoɔ-twins and tani-one of the twins), reference is made to immigrants
32
2 Inter-party insults in political discourse in Ghana: A critical discourse analysis
from the Northern part of Ghana; others have the understanding that it is an insult
which makes reference to people from the northern part of Ghana as “dirty people”. The
common idea out of the two meanings is that it is not a good term to use for Northerners.
Interestingly, this term surfaced on the front page of an NDC newspaper “tani can’t
govern us”. Take note of the fact that the 2012 presidential candidate of the NDC and the
current President is a Northerner. Also, it is important to recall that the NPP has long
been perceived as an Akan or Ashanti dominated party, and historically Northerners
were considered as servants of the Akans. Putting all these facts in context, the pro-NDC
newspaper pitches an old ideological battle between the ingroup and the outgroup. The
outgroup is presented negatively as insulting and marginalizing people from the North,
consequently raising ethnic tension between people from the North and the outgroup,
that is, turning the people from the North against the NPP. According to van Dijk (2001),
ethnic prejudice and ideologies are not innate; rather, they are acquired and learned
through communication, that is, through text and talk.
Secondly, with regard to detailed description, pro-NPP newspapers employed detailed
positive descriptions to describe ingroup members as well as non-politicians insulting
the outgroup. For example, the Daily Searchlight paper described Mr. Mohammed Ameen
Adams, who accused the deputy energy minster, Alhaji Inusah Fuseini of being economical with truth concerning the energy crisis in the country as “an energy expert or
economist”. Similarly, the Daily Guide paper described Fiifi Banson, a broadcast journalist, as “an award winning Ghanaian broadcaster” when he criticized the outgroup for not
sending people who can make an intellectual debates. The paper, however, referred to
former President Rawlings as “founder of the ruling NDC” when he insulted the party he
founded. Negative descriptions were also used to refer to outgroup members for insulting the ingroup. For example, the former President Rawlings was described by the Daily
Guide paper as “see no evil” for calling former President Kufour an “autocratic thief”.
NDC newspapers, on the other hand, employed honorifics and official titles to describe ingroup members who insulted the outgroup. For example, the Voice newspaper
described Hamza Abugri, the Bantama constituency organizer of the NDC, as “honorable” for insulting the chairman of the NPP as “ignorant”. This is very interesting because “honorable” is a title given to Ministers of state, MPs, Metropolitan, Municipal and
District Chief executives (MMDCE), and assembly members but not leaders of political
parties. Members of the outgroup who insulted the outgroup were given more detailed
descriptions such as “leading member of NPP” (Daily Post), “senior member of NPP” (Radio Gold), and “a stalwart of the opposition” (Informer), to report such insults as credible.
Negative descriptions were also used to refer to outgroup members’ insults targeted at
the in-group. For example, “genocide MP” and “loose talking MP” (The Catalyst).
van Dijk (1995) aptly states that one of the structures used to present the ingroup positively is detailed description. This is supported by Blommaert (2005), who points out
that members of the ingroup employ indexical meaning such as terms of politeness to
elevate them to a particular social status. It is however important to note that, in this
paper, it has been revealed that positive descriptions were used to describe the outgroup
members who insulted their own party to portray to readers that the source of the insult
33
Emmanuel Amo Ofori
or information is credible and authentic. Though the opposing group members are presented positively, it gives an impression to readers that there is confusion at the camp
of the outgroup, which eventually presents them negatively.
In sum, the two important ideological structures used were negative lexicalization
and predication as well as detailed description. A detailed analysis of these structures
revealed group polarization between the pro-NPP and NDC newspapers.
5 Conclusion
Utilizing Fairclough’s textual analysis and van Dijk’s concept of ideological square, this
paper has revealed the Us/Them representation of insults between the NPP and NDC
newspapers. That is to say, the actions of the ingroup were presented positively while
those of the outgroup were presented negatively. The ideological differences and political
spin in the representation of insults showed a clear group polarization between NPP and
NDC newspapers. The various lexicalizations and predications employed by both NPP
and NDC newspapers revealed the Us/Them dichotomy between these two dominant political parties in Ghana. Using the authorial voice and other sources, the presuppositions
also showed the ideological differences between the papers. Finally, verbal processes employed by NPP and NDC papers clearly manifested the positive ingroup representation
in NDC papers as compared to NPP paper. Negative other representation was a common
trait in both papers.
References
Agyekum, Kofi. 2010. Akan verbal taboos in the context of ethnography of communication.
Accra: Ghana Universities Press.
Althusser, Louis. 1971. Ideology and ideological state apparatuses. In Louis Althusser
(ed.), Lenin and philosophy and other essays, 121–173. London: New Left Books.
Bayram, Fatih. 2010. Ideology and political discourse: A critical discourse analysis of
Edorgan’s political speech. ORECLS 7. 23–40.
Blommaert, Jan. 2005. Discourse: A critical introduction. Cambridge: Cambridge University Press.
Cameron, Deborah. 2012. Verbal hygiene. New York: Routledge.
Chen, Lily. 2005. Transitivity in media texts: Negative verbal process sub-functions and
narrator bias. IRAL 43. 33–51.
Fairclough, Norman. 1989. Language and power. New York: Longman.
Fairclough, Norman. 1995. Media discourse. London: Arnold.
Fowler, Roger. 1987. The intervention of the media in the reproduction of power. In Myriam Díaz-Diocaretz, Teun A. van Dijk & Iris M. Zavala (eds.), Approaches to discourse,
poetics and psychiatry, 67–80. Amsterdam: Benjamins.
Gramsci, Antonio. 1971. Selections from the prison notebooks of Antonio Gramsci. New
York: International Publishers.
34
2 Inter-party insults in political discourse in Ghana: A critical discourse analysis
Halliday, Michael. 1985. An introduction to functional grammar. London: Arnold.
Hasty, Jennifer. 2005. The press and political culture in Ghana.
Marfo, Samuel. 2014. Thinking of peace, democracy and politics of insults in Ghana: The
paradox of freedom and culture of violence. Online Journal of African Affairs 3. 139–
144.
Ninsin, Kwame. 2006. Political parties and political participation in Ghana. Accra: Konrad
Adenauer Foundation.
Owusu, William. 2012. The ghanaian media landscape: How unethical practices of journalists undermine progress. Oxford: no publisher. Unpublished Masters thesis.
Reisigl, Martin & Ruth Wodak. 2001. Discourse and discrimination: Rhetorics of racism
and antisemitism. London: Routledge.
Richardson, John. 2007. Analyzing newspapers: An approach from critical discourse analysis. Houndmills: Palgrave.
van Dijk, Teun. 1995. Discourse analysis as ideology analysis. In Christina Schaffner &
Anita Wenden (eds.), Language and peace, 17–33. Dartmouth: Dartmouth Publishing
Company Limited.
van Dijk, Teun. 1998. Opinions and ideologies in the press. In Allan Bell & Peter Garret
(eds.), Approaches to media discourse, 21–63. Oxford: Blackwell.
van Dijk, Teun. 2001. Critical discourse analysis. In Deborah Schiffrin, Deborah Tannen &
Heidi Hamilton (eds.), The handbook of discourse analysis, 352–371. Oxford: Blackwell.
Wodak, Ruth. 2001. What CDA is about - A summary of its history, important concepts,
and its developments. In Ruth Wodak & Michael Meyer (eds.), Methods of critical discourse analysis, 1–13. London: Sage.
Wodak, Ruth. 2007. Pragmatics and critical discourse analysis: A cross-disciplinary inquiry. Pragmatics & Cognition 15(1). 203–225.
Wodak, Ruth & Michael Meyer. 2009. Critical discourse analysis: History, agenda, theory,
and methodology. In Ruth Wodak & Michael Meyer (eds.), Methods of critical discourse
analysis, 1–33. London: Sage.
Wood, Linda A & Rolf O Kroger. 2000. Doing discourse analysis: Methods for studying
action in talk and text. London: Sage.
Yiannis, Gabriel. 1998. An introduction to the social psychology of insults in organizations. Human Relations 51. 1329–1354.
35
Chapter 3
Classification of Guébie within Kru
Hannah Leigh Sande
Guébie, a Kru language spoken in Côte d’Ivoire, is currently doubly classified within Eastern Kru according to Ethnologue (Lewis et al. 2013). It is listed as a dialect of two distinct
subgroups, Bété and Dida. This double classification is clearly problematic, and this paper
provides the initial work towards addressing the correct classification of the language. Here
I compare the phonological and syntactic properties of Guébie with surrounding Bété and
Dida languages in order to determine its relatedness to each subgroup. I conclude that Guébie is more closely related to Vata, a Dida language, than to Bété.
1 Introduction
Kru is a branch of Niger-Congo languages spoken in Côte d’Ivoire and Liberia. There
are two major subdivisions of Kru languages: Eastern and Western Kru (Marchese 1979);
however, there has been very little work done on the internal classification of those two
branches. Many ‘languages’ classified as either Eastern or Western Kru are in fact subgroups of related languages. For this reason, many Kru languages are grouped together
as a single language for classification purposes, sometimes for empirical reasons and
sometimes for political or geographic ones, leading to linguistic misclassifications and
inadequate descriptions of individual Kru varieties. In this paper I address one such case,
that of Guébie, an Eastern Kru language spoken in southwest Côte d’Ivoire. I attempt to
provide an initial classification of this particular Kru language.
Guébie is a particularly interesting case of misclassification. It is currently classified
twice in Ethnologue (Lewis et al. 2013),1 and I argue here that neither classification is
accurate. Lewis et al. (2013) calls Guébie both a dialect of Bété-Gagnoa (btg), and an
alternative name for Dida-Lakota (dic). Bété-Gagnoa has 150,000 speakers according to
Ethnologue (Lewis et al. 2013), and Dida-Lakota (dic) has 94,000 speakers. Based on my
own observations during fieldtrips along with estimates from consultants, I estimate that
Guébie, the language in question, has only 7,000 remaining speakers. Thus, classifying
Guébie as a dialect of Bété-Gagnoa or an alternative name for Dida-Lakota is incredibly
misleading in terms of the number of remaining speakers and vitality of the language.
1 The revised version of this chapter was completed in 2015. The version of Ethnologue cited throughout this
paper is the 2013 edition, which was the most recent at the time of writing.
Hannah Leigh Sande. Classification of Guébie within Kru. In Jason Kandybowicz, Travis Major,
Harold Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected papers
from the 45th Annual Conference on African Linguistics, 37–49. Berlin: Language Science Press.
DOI:10.5281/zenodo.1251712
Hannah Leigh Sande
The goal of this paper is to determine the appropriate classification of Guébie and
advocate that it be classified separately from both Bété-Gagnoa and Dida-Lakota. I will
demonstrate based on original Guébie data that Guébie is more closely related to Vata,
a dialect of Dida-Lakota, than to Gbadi, a dialect of Bété-Gagnoa. I choose these two
particular varieties as standards of comparison because there is more data available for
Vata and Gbadi than for other Dida and Bété languages. I conclude based on this initial
study that Guébie is more closely related to Dida languages than Bété ones, and further,
that it is distinct from Vata (Dida-Lakota), thus it should be classified as a distinct Dida
language.
I begin in §2 with background information on Guébie and its current classification
within Kru. I turn in §3 to the methodology used here to determine relatedness of languages. In §4 I discuss the somewhat limited phonetic and phonological evidence that
Guébie is more closely related to Dida-Lakota than to Bété-Gagnoa, and in §5 I discuss
the more readily available syntactic data which supports the conclusion that Guébie is a
Dida language, but is distinct from Dida-Lakota. I conclude in §6 with the implications
and conclusions of the present study, arguing that Guébie be classified as distinct from
both Bété-Gagnoa and Dida-Lakota, contra the current Ethnologue classification (Lewis
et al. 2013).
2 Background: The current classification of Guébie
The current literature agrees that Kru is a branch of Niger-Congo (cf. Marchese Zogbo
2012). However, there have been many other contradictory proposals in the past. There
have been claims that Kru is related to Gur (Vogler 1974), Kwa (Greenberg 1963), and
Mande (Bennett & Sterk 1977). Welmers (1977) put forth the proposal that Kru is a distinct
branch of Niger-Congo, and that has been corroborated by Marchese (1979) and later
work. There are two major branches of Kru languages, Eastern and Western, which are
geographically split near the country boarder of Côte d’Ivoire and Liberia. Guébie is
an undescribed Eastern Kru language spoken in seven small villages in southwest Côte
d’Ivoire about 30 kilometers southwest of Gagnoa and 30 kilometers east of Lakota. The
largest of these villages, Gnagbodougnoa, has a population of 1000, about two thirds of
whom are Guébie-speaking. The remaining third are Lobi speakers who were displaced
during the national crisis of Côte d’Ivoire in the 1990s. Since Guébie is spoken in just
seven villages, the largest of which has a population of 1000, I estimate that Guébie has
at most 7000 speakers.
The data presented in this paper comes from original work with native speakers of
Guébie. For eight months I worked with a native speaker temporarily living in Berkeley,
California. This was followed by fieldwork in Gnagbodougnoa, Côte d’Ivoire in the summer of 2014. My consultants are five in number, include both men and women, and range
in age from 19–76. This paper presents the first published documentation and description
of Guébie.
Guébie villages are situated amidst a dense rainforest where temperatures are high
year-round and there are two rainy seasons, one in June and the other in December.
38
3 Classification of Guébie within Kru
The community of Guébie speakers are subsistance farmers who grow cassava, rice, and
plantains. Rarely are these plants farmed for profit. A small percentage of Guébie people farm and sell cacao and coffee. Only one third of children attend school, while the
others work on their family’s farm. There is only one known monolingual speaker, my
consultant Serikpa Emil, who is 76 years old. The rest of the Guébie-speaking population is bilingual in French. Children are still learning Guébie, however within a single
generation, I predict that children will no longer learn Guébie as a first language.
The Eastern Kru family tree in Figure 1 is adapted from Ethnologue (Lewis et al. 2013).
Eastern Kru
Bakwe-Wane
Bété
Eastern
Gagnoa
Dida
Western
Kwadia
Lakota
Kouya Daloa Guiberoua Godié Vata/Guébie Dida-Yocoboué Neyo
Guébie Gbadi
Figure 1: Eastern Kru family tree
Guébie, bolded in the tree above, is currently classified twice, as part of two distince
Eastern Kru varieties. Due to geographic, historical, and possibly ethnic reasons, Guébie
has been called a dialect of Bété-Gagnoa, spoken by about 150,000 people (Lewis et al.
2013). Due to linguistic similarity, Guébie is also listed as a second name for Dida-Lakota,
an Eastern Kru language spoken by about 94,000 people in the city of Lakota, east of the
Guébie-speaking area.
3 Methodology
The most widely accepted means of establishing genetic relationship between languages
is the Comparative Method (cf. Bloomfield 1933 chapter 18, among others), which determines whether sound correspondences across languages are regular, thus the result of
regular sound change. For the Comparative Method to be of use requires dictionaries or
lists of lexical items in the languages in question to be used for comparison.
Ideally, we would compare sound correspondences across cognates in the lexicon of
Guébie with the lexicon of Bété-Gbadi and Vata (Dida-Lakota) in order to determine
relatedness. However, there has not been enough thorough documentation across Bété
and Dida languages for comparing sound correspondences to be informative. The available resources for Guébie include only my own data elicted from September 2013 to July
2014. For Bété-Gbadi, there is a dictionary written by a native speaker linguist (Zogbo
2005). For Vata there is a small list of lexical items compiled from various works on
the phonology and syntax of the language (Marchese 1979; Kaye 1981; Koopman 1984).
39
Hannah Leigh Sande
Amongst these resources, I have found fewer than 100 cognates to compare across the
three languages. This number of cognates is not enough to determine relatedness based
on sound change; however, I will present what minimal data exists in §4.
For each of the languages in question, Guébie, Bété-Gbadi, and Vata, there is some
not insignificant amount of syntactic description in the literature. Basing genetic relationship on syntactic correspondences is less widely accepted than on sound correspondences and morphological paradigms (Garrett 2012; Ringe & Eska 2014, and citations
therein). However, since there are no available lexicon lists for most Kru languages, I
propose that we examine syntactic features of Bété and Dida languages, comparing them
with Guébie to determine how these languages are related.
Lexical items are likely to be borrowed heavily from neighboring languages, and morphosyntactic correspondences are more likely to be conservative or undergo change less
rapidly over time. Thus, if we can find convincing syntactic similarities between two languages, A and B (say, Guébie and Vata), but not between two others, A and C (say, Guébie
and Bété-Gbadi), we may be able to tentatively say that the first two languages, A and
B, are more closely related than the latter two. Comparing morphosyntactic features of
languages has been crucial in determining the relatedness of Proto-Anatolian languages
(Melchert 2013), and the place of Armenian within Indo-European (Hübschmann 1875).
Thus, it is possible, however rare, for morphosyntactic similarity to influence decisions
about language classification.
In the following two sections I compare the linguistic properties of Guébie with its
two geographically closest documented neighbors, Bété-Gbadi and Vata. Bété-Gbadi is
spoken in just north of the Guébie-speaking villages. The Bété-Gbadi data here comes
from Marchese (1979); Zogbo (2005); Koopman (1984). Vata is a Dida language spoken
east of the Guébie-speaking villages. The data here comes from Marchese (1979); Kaye
(1981; 1982); Koopman (1984). Vata is a dialect of Dida-Lakota spoken slightly northeast
of Lakota. Ideally we would compare Guébie with Dida-Lakota spoken in the city of
Lakota; however, there is very little work on the Dida spoken in Lakota. Vata is slightly
geographically further from Guébie; however, since it is a well documented and closely
related to Dida-Lakota, I compare Vata, not the Dida of Lakota, with Guébie.
I demonstrate throughout the remainder of this paper that the linguistic properties
of Guébie and Vata are too similar to be due to chance, and that Guébie is less closely
related to Bété-Gbadi than to Vata. The words and sounds that do correspond in Guébie
and Bété must be due to borrowing, or must have been present in Proto-Eastern-Kru.
There is very little evidence that Bété and Guébie underwent any regular changes that
other Eastern Kru languages did not undergo. Any assumptions made here about the
features of Proto-Kru or Proto-Eastern-Kru come from Marchese (1979) and Marchese
Zogbo (2012).
4 Phonetic and phonological evidence
There is very little, if any, phonetic or phonological evidence that can concretely determine the classification of Guébie as Bété or Dida. I present here some basic similarities
40
3 Classification of Guébie within Kru
and differences between Bété-Gbadi and Guébie, and Vata and Guébie.
4.1 Tone
The data in (1) shows that there are four contrastive tonal heights in Guébie. The words in
(1a) are segmentally identical and are distinguished only by the height of their level tone.
The data in (2) shows that Guébie also makes use of contrastive contour tones. There are
two distinct rising tones and three distinct falling tones in Guébie.
(1)
Four contrastive tone heights in Guébie
a. ko1 ‘line/row’ ko2 ‘pestle’
b.
no2 ‘beverage’
1
c. ɟe
‘egg’
ko3
‘skin’
ɟe3
‘star’
ko4
no4
ɟe4
‘cadavre’
‘woman’
‘number’
(2) Contour examples
ɟa31 ‘coconuts’
vɔ13 ‘horns’
su2
‘tree’
su13 ‘to shove’
There are four contrastive tonal heights in Guébie. There are also four contrastive tonal
heights in Vata (Marchese 1979; Kaye 1981; Koopman 1984). It is controversial whether
there are three or four contrastive tonal heights in Bété-Gbadi. Zogbo (2005)’s dictionary
lists four contrastive tones for Bété-Gbadi; however, Marchese (1979; 1989) says that the
four tonal heights posited for Proto-Kru have collapsed into three heights in Bété. Tones
throughout this paper are marked with numbers 1–4, where 4 is the highest tone and 1 is
the lowest. A dot between tones separates syllables, and two numbers within a syllable
signifies a contour tone.
4.2 Vowels
There are ten contrastive vowels in Guébie, distinguished by height, backness, rounding
and ATR value. /ə/ is the +ATR counterpart of /a/ in Guébie.
41
Hannah Leigh Sande
(3)
Guébie vowel inventory
u
i
ʊ
ɪ
e
o
ә
ε
ɔ
a
There are also ten contrastive vowels in Vata, identical to those in the vowel inventory
in the chart above. There are only seven contrastive vowels in Bété-Gbadi. Bété-Gbadi
lacks an ATR distinction between the mid and low vowels. Marchese Zogbo (2012) posits
seven contrastive vowels for Proto-Kru, which means that Vata and Guébie each separately underwent vowel splits resulting in a larger vowel inventory. Alternatively, Vata
and Guébie share a common ancestor that Bété-Gbadi does not share, and that ancestor
underwent regular phoneme splits, adding three vowels to the inventory. This ten-vowel
inventory was then passed down to both Vata and Guébie. Without more information
we cannot say for sure which path of development of these vowel systems is the correct
one.
There are marginal nasal vowels in Guébie. The three vowels [ɛ̃, ɔ̃, ã] are found in a
just a handful of words. Proto-Eastern-Kru did not have nasal vowels. However, nearby
Kwa and Mande languages, frequently in contact with Kru languages, have these same
three nasal vowels. This contact could have resulted in the borrowing of Kwa and Mande
words into Guébie, so that the borrowed words but no native words contain nasal vowels
in Guébie. The three Guébie words that I have found containing nasal vowels are given
in (4).
(4)
Guébie nasal vowels
‘spine’
a. kãɔ̃4.2
b. ɟiɛ̃3.1
‘sea’
c. kpãɛ̃4.4 ‘very, a lot’
In Bété-Gbadi, nasal vowels are also found in a few, likely non-native, words. One
of these is the same word, kpãɛ̃ ‘very’, that contains a nasal vowel in Guébie. Because
both Guébie and Bété-Gbadi have had contact with surrounding Kwa languages in the
past, it seems likely that both languages borrowed words containing nasal vowels from
those Kwa languages. Alternatively, one of the two, Guébie or Bété-Gbadi, could have
borrowed the words in (4) and these words could have in turn been borrowed into the
other. Either way, it seems unlikely that nasal vowels were an innovation via regular
sound change shared by Guébie and Bété; borrowing seems like a more plausible option
because there are so few words in each language that contain nasal vowels, and the
words that do have nasal vowels overlap in Guébie and Bété-Gbadi.
42
3 Classification of Guébie within Kru
Guébie has pervasive ATR harmony from roots to suffixes. Vata shares the same ATR
harmony process, though Bété-Guébie does not (Koopman 1984). Guébie ATR harmony
can be seen in (5) where the causative morpheme is a low vowel suffix on verbs. Verb
roots that contain +ATR vowels take the +ATR causative suffix [-ə] while those containing -ATR vowels in the root take the -ATR causative suffix [-a]. The same process can
be seen with the Vata definite marker, where +ATR root vowels result in +ATR suffix
vowels. There is rounding harmony in the Vata data that does not occur in Guébie. The
Gbadi data shows that +ATR roots do not result in +ATR suffixes. This is likely a factor
of the limited vowel inventory of Gbadi, however, where only high vowels show an ATR
distinction. There are no suffixes in Gbadi that contain high vowels, so we can not say
for certain whether or not there would be ATR harmony between roots and suffixes in
high vowels in Gbadi.
(5) Vowel harmony data
Guébie ci-ə3.1
‘to cause to learn’
Vata
sle-e2.3 ‘the house’
Gbadi
li-a2.2
‘to cause to eat’
jɛ-a3.2
ɡbʊ-ɔ2.3
jue-a4.4.2
‘to cause to dance’
‘the cause’
‘the children’
Without a high-vowel suffix in Bété-Gbadi, we cannot say for certain whether all three
languages have ATR harmony, or whether only Vata and Guébie share this ATR harmony
process. If the latter is true, either this harmony process arose separately in Vata and
Guébie, or it was lost relatively recently in Gbadi. Alternatively, Vata and Guébie share
a common ancestor that Bété does not share, and that ancestor acquired a harmony
process that Bété did not. More data is needed to know for certain.
4.3 Consonants
There is too little lexical data available from documented Bété and Dida languages to
show regular sound changes in consonants that led from a Proto-language to the currently spoken languages.
4.4 Summary
Until further data is collected, the existing phonological data on these languages do not
tell us much about their genetic relationship. What we can conclude from the above is
that there are no known shared changes between only Guébie and Bété-Gbadi. All of
the features that Guébie shares with Bété are also present in Vata, or there is evidence
that those features a result of borrowing (nasal vowels). In the following section I turn
to syntactic evidence of relatedness.
5 Syntactic evidence
In this section I compare certain syntactic properties of Guébie with those of Bété-Gbadi
and Vata (Dida). I am limited by the specific syntactic properties that have been described
43
Hannah Leigh Sande
for all three languages in question, though there are numerous syntactic features of Guébie that more closely resemble Vata than Bété-Gbadi and other Bété dialects.
Before describing the differences, it is important to note that all three languages share
S AUX O V alternating with SVO word order. I turn now to those properties that are not
shared amongst all three languages.
5.1 Aspect marking
The two major aspectual distinctions in Guébie, imperfective and perfect, are distinguished by tone. All imperfective verbs have tone one step lower on the four-tone scale
than the corresponding perfect verb, (6).
(6) Perfective vs Imperfective in Guébie
ɟa31
a. ɔ3 li2
3.sg eat.impf coconuts
‘He eats coconuts’
ɟa31
b. ɔ3 li3
3.sg eat.perf coconuts
‘He ate coconuts (recently)’
Vata (and Dida-Lakota, Kaye 1982) also distinguishes imperfective from perfective aspect with tone.
Bété-Gbadi, on the other hand, distinguishes imperfective from perfective aspect with
auxiliary particles, and the verb surfaces finally: S AUX O V.
(7)
Comparing aspect distinctions
Language Perfective Imperfective
Guébie
Tonal
Tonal
Vata
Tonal
Tonal
Gbadi
Particle
Particle
This difference means that in Proto-Eastern-Kru there was either an aspect-marking
auxiliary that was reduced to a tonal morpheme in Guébie and Vata, or there was historically a tonal morpheme that was replaced by an auxiliary in Bété-Gbadi, but not in all
Eastern Kru languages.
5.2 Causation
There are two methods of adding a causative meaning to a verb in Guébie. One of these
is suffixal, shown in (8).
(8)
44
Guébie suffixal causative
a. ci31 ‘to learn’ ci-ə3.1
b. jɛ3
‘to dance’ jɛ-a3.2
‘to cause to learn, to teach’
‘to cause to dance’
3 Classification of Guébie within Kru
The second means of causativization in Guébie is with a clausal construction meaning
literally “X speaks and Y Zs,” where X is the causer, and Y is the subject of the verb Z.
(9) Clausal causative in Guébie
kɔɡʊlɪɲɔ-wa4.2.2.2.3 gba2 ne4 ju-wa4.4 ɔ3 li2
speak and boy-def 3.sg eat.perf
farmer-def
‘The farmer is making the boy eat’
The suffixal causative in (8) is present in both Vata and Bété-Gbadi as well. However,
the clausal causative is only found in Vata and Guébie, not Bété-Gbadi.
(10) Clausal causative in Vata
n3 ɡba2 le3 yɔ-ɔ3.3 li2
I speak and boy-def eat
‘I made the boy eat.’
It is possible that the clausal causative construction was borrowed into Guébie from
Vata or vice versa; however, it is also possible that the two share a common ancestor that
Bété does not share, and the clausal causative was innovated in that ancestor language,
inherited into both Guébie and Vata. All we can say is that there is no commonality
between Bété and Guébie that Vata does not also share.
5.3 WH-questions
Wh-questions in Guébie are formed with a clause-final question marker and a clauseinitial question word meaning literally ‘person, place, thing’ for ‘who, where, what,’ respectively.
(11) Wh-questions in Guébie
a. ɲɔkpa3.3 touri1.1.3 ji3 lɛtrɪ3.2 kɔpa3.23 na3
q
Touri.name will letter send
who
‘To whom will Touri send a letter?’
kɔpa3.23 na3
b. bɛba2.2 touri1.1.3 ji3 ɟaci2.2
q
what Touri.name will Djatchi.name send
‘What will Touri send to Djatchi?’
According to (Koopman 1984: 87), Wh-questions are formed in Bété-Gbadi with an
initial Wh-word and a clause-medial question marker, while in Vata they are formed
with an initial Wh-word and a final question marker. This means that once again the Vata
construction (12) is identical to the Guébie construction, while the Bété construction is
distinct. The chart in (13) shows the similarity between the Guébie and Vata but not the
Guébie and Bété Wh-construction.
45
Hannah Leigh Sande
(12)
Wh-questions in Vata
alɔ1.4 ɔ1 le2 saka3.4 la1
wh
who he eat rice
‘Who eats rice?’
(13)
Comparing Wh-constructions
Language Initial Medial
Guébie
Wh
–
Vata
Wh
–
Gbadi
Wh
+q
Final
+q
+q
–
Much like the causative construction in the previous section, we see a similarity between Guébie and Vata that could be the result of borrowing or common inheritance.
More data, prefereably historical data, is needed to know for certain.
5.4 Gerunds
There are two types of gerund formation in Guébie. One form of the gerund is formed
by reduplicating the verb and adding the suffix /-je/. The other is formed by adding the
suffix /-li/ to the verb2 . The former gerund construction is found in Bété-Gbadi and not
Vata, while the latter gerund construction is found in Vata but not Bété-Gbadi (Koopman
1984).
(14)
Gerunds in Guébie
a. saka3.3 la2 li-li-je2.2.1
gen eat-eat-nom
rice
‘Rice-eating’
b. saka3.3 la2 pi-li3.1
gen cook-nom
rice
‘Rice-preparing’
(15)
Gerunds in Vata
saka3.4 la2 pi-li2.1
gen cook-nom
rice
‘Rice-preparing’
(16)
Gerunds in Bété-Gbadi
li-li-je2.2.1
eat-eat-nom
‘Eating’
The Guébie construction in (14b) is identical to the Vata construction in (15). Likewise,
the Guébie construction in (14a) is shared by Bété-Gbadi, as shown in (16). Without knowing which gerund formation strategy was present in the proto language, we cannot make
2 The
two gerund formation strategies described here have distinct distributions that I cannot yet cleanly
define.
46
3 Classification of Guébie within Kru
any conclusions about whether each of these gerund constructions was inherited or borrowed. Without further information, gerunds do not shed light on the classification of
Guébie.
5.5 Summary
We have seen that Guébie syntax is similar to Vata in aspect marking, causation, and
wh-question formation. It shares ones gerund construction with Vata and another with
Gbadi. However, there are very few other syntactic similarities between Guébie and
Bété-Gbadi. Vata and Guébie share further syntactic properties not presented here for
purposes of space. For example, they have identical applicative constructions, similar
negation marking, and they both can undergo long-distance wh-movement while BétéGbadi cannot.
The table in (17) below summarizes the phonological and syntactic features compared
here in Guébie, Vata, and Bété-Gbadi.
(17) Summary table
Feature
# of tones
# of vowels
Aspect
Clausal causation
Wh-question particle
Gerunds
Bété
3
7
particle
–
medial
verb-verb-je
Guébie
4
10
tone
yes
final
verb-verb-je, verb-li
Vata
4
10
tone
yes
final
verb-li
The Vata and Guébie columns above are nearly identical. The two share the same number of contrastive tones and the same vowel inventory. They both mark aspect distinctions with tone, they have clausal causative constructions, they have final Wh-particles,
and they share a gerund construction. Guébie shares a gerund construction with BétéGbadi but in all other respects there are key differences between the two.
Based on the limited phonological and syntactic data available for Guébie, Vata, and
Bété-Gbadi, it seems that Guébie shares far more features with Vata, a Dida language,
than with Bété-Gbadi, a Bété language. In the following section I argue that we should
classify Guébie as a distint language in the Dida subgroup of Eastern Kru.
6 Implications and conclusions
We have seen evidence from the phonological and prosodic systems, and the morphosyntax of Guébie, that it resembles Vata, a Dida language, more closely than Bété-Gbadi, a
Bété langauge. Further research on Bété and Dida languages will allow for lexical and
sound-correspondence comparison as well, which will confirm or deny the claims made
here.
I return here to the problem defined in §1 of this paper: where within in Eastern Kru
should Guébie be classified. We saw that in Ethnologue (Lewis et al. 2013) Guébie is
47
Hannah Leigh Sande
currently classified twice, once as a dialect of Bété-Gagnoa, and once as an alternative
name for Dida-Lakota. Dida and Bété are distinct subgroups of Eastern Kru with distinct histories and distinct lingusitic features. The goal of this paper is to make an initial
step towards determining whether Guébie is a Bété or Dida language. No firm conclusions can be made without further research; however, Guébie is culturally distinct from
both of its neighboring Bété and Dida languages, and I argue here that it is also linguistically distinct. Guébie should be classified as a distinct language, and based on the
data presented throughout this paper, I tentatively conclude that Guébie is a Dida language, closely related to Dida-Lakota and Vata, though not synonymous with either of
them. This conclusion is based on the fact that there are many phonological and syntactic features shared between Guébie and Vata that are not present in Bété-Gbadi. It
seems unlikely that all of the similarities between Guébie and Vata are due to chance or
borrowing; thus, inheritence is a more plausible history.
Eastern Kru
Bakwe-Wane
Bété
Eastern
Dida
Western
Lakota
Kwadia
Dida-Yocoboué Neyo
Gagnoa Kouya Daloa Guiberoua Godié Vata Guébie
Gbadi
Figure 2: My proposed classification of Guébie within Kru
The tree in Figure 2 shows my proposed classification of Guébie, as a Dida-Lakota
language related to but distinct from Vata. Crucially, I claim that Guébie is not a dialect
of Bété-Gagnoa, and in fact it is not a Bété language at all.
Far more data is needed to prove the above classification and the further internal classification of Kru; however, this paper provides and initial step towards a more detailed
understanding of the Kru languages and how they are related to each other. Arguing for
the classification of Guébie at least provides a strong argument that can be argued for
or against in future work when further data becomes available.
Finer grained classification than that shown in the tree above will require extensive
further research. Comparative work such as Kaye (1982), “Les dialects dida,” is a start
toward this kind of comparative research in Kru.
48
3 Classification of Guébie within Kru
References
Bennett, Patrick R. & Jan P Sterk. 1977. South central niger-congo: A reclassification.
Studies in African linguistics 8. 241–273.
Bloomfield, Leonard. 1933. Language. New York: Henry Holt.
Garrett, Andrew. 2012. The historical syntax problem: Reanalysis and directionality. In
Dianne Jonas, John Whitman & Andrew Garrett (eds.), Grammatical change: Origins,
nature, outcomes, 52–72. Oxford: Oxford University Press.
Greenberg, Joseph H. 1963. Some universals of grammar with particular reference to the
order of meaningful elements. In Joseph Greenberg (ed.), Universals of language, 73–
113. Cambridge: MIT Press.
Hübschmann, Heinrich. 1875. Uber die stellung des armenischen im kreise der indogermanischen Sprachen. Zeitschrift für Vergleichende Sprachforschung 23. 5–42.
Kaye, Jonathan D. 1981. Las ́election des formes pronominales en vata. Revue qúebecoise
de linguistique 11. 117–135.
Kaye, Jonathan D. 1982. Les dialectes dida. In Jonathan D. Kaye, Hilda Koopman & Dominique Sportiche (eds.), Projet sur les langues Kru, 233–295. Montreal: Quebec University.
Koopman, Hilda. 1984. The syntax of verbs. Dordrecht: Foris Publications.
Lewis, M. Paul, F. Gary Simons & Charles D. Fenning (eds.). 2013. Ethnologue: Languages
of the world. 17th edn. Dallas: SIL International.
Marchese Zogbo, Lynell. 2012. Kru revisited, Kru revealed.
Marchese, Lynell. 1979. Atlas linguistique kru. Abidjan: ILA.
Marchese, Lynell. 1989. Kru. In Bendor Samuel (ed.), The Niger-Congo languages, 119–213.
Lanham: University Press of America Inc.
Melchert, H. Craig. 2013. The position of Anatolian. In Michael Weiss & Andrew Garrett
(eds.), Handbook of Indo-European studies, 85–163. Oxford: Oxford University Press.
Ringe, Don & Joseph F. Eska. 2014. Historical linguistics: Toward a twenty-first century
integration. Cambridge: Cambridge University Press.
Vogler, Pierre. 1974. Le problème linguistique kru, éléments de comparaison. Journal de
la Socíeté des Africanistes 44. 147–176.
Welmers, William E. 1977. The Kru languages: A progress report. In Language and linguistic problems in Africa: Proceedings of the Seventh Conference on African Linguistics,
353–62. Columbia, SC: Hornbeam.
Zogbo, Raymond Gnoleba. 2005. Dictionnaire bété-français. Abidjan: Editions du CERAP.
49
Chapter 4
What about Southern African story
grammar? Promoting language specific
macrostructures in educational settings
Heike Tappe
University of KwaZulu-Natal, Howard College Campus, Linguistics Programme
The current research is motivated by the assumption that the “canonical” story grammar
(Stein & Glenn 1979) which is widely held to reflect universal story organisation (e.g. Mandler et al. 1980) may not be fully suited to assess the narrative potential of children with primary languages of non-European origin (e.g. Souto-Manning 2013). It thus juxtaposes the
supposedly universal “canonical” story grammar with the idea of an as yet under-researched
Southern African story grammar, which may represent common macro-structural features
in Southern African storytelling.
The analysis of Southern African folktales from Lesotho, South Africa, Malawi and Zambia
as well as narrative retellings produced by multilingual children in Malawi, whose dominant
languages are Chichewa and English (Tappe & Hara 2013; Hara 2014), and in South Africa,
whose dominant languages are isiZulu and English, provides evidence that Southern African
narratives systematically deviate from the allegedly universal “canonical” story grammar
(Anderson & Evans 1996; Stein & Glenn 1979). In both Southern African countries currently
under investigation the “canonical” story grammar (CSG) is, however, promoted through
the language of teaching and learning (English) and determines the assessment instruments
used to measure the children’s narrative skills.
Against this background the emancipation of Southern African ways of storytelling might
be a crucial step in the fight against literacy in Southern African countries.
1 Introduction
Globally, education is conceptualized as a tool that serves the improvement of social
justice and economic advancement. Accordingly, equal access to the full spectrum of
the educational system was one major concern during protests against colonisation in
all Southern African countries as the redistribution in social justice was widely equated
with access to the kind of education that many term “quality education” (e.g. Tikly &
Heike Tappe. What about Southern African story grammar? Promoting language specific macrostructures in educational settings. In Jason Kandybowicz, Travis Major, Harold
Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected papers from
the 45th Annual Conference on African Linguistics, 51–64. Berlin: Language Science Press.
DOI:10.5281/zenodo.1251714
Heike Tappe
Barrett 2011). Quality education can be conceptualised as education that enables learners
to become “economically productive, healthy, secure and active citizens” (Tikly & Barrett
2011: 9).
However, in post-colonial Southern Africa “economically productive, healthy, secure
and active citizens” are still predominantly of European descent, or have had access to
education that is based on European curricula; hence the term “quality education” has
become largely synonymous with European curricula taught in a European language.
A considerable number of empirical studies (e.g. Orman 2008; Nomlomo 2010; Nkosana 2011; Evans & Cleghorn 2014) provide evidence for this assessment as they observe
that speakers of African languages constantly face the conundrum of cherishing their
native language(s) on the one hand but wanting their children to be speakers of a European language (even if it is at the expense of their home language) on the other hand.
Consequently, it has become a generally acknowledged common practise for schools
in Southern Africa to offer instruction in a locally spoken, indigenous language for the
first three to five years of schooling before an international language is introduced as a
medium of instruction (e.g. Bloch 2006: 3).
Tragically, the negligence of children’s native language(s) leads to great number of
almost insurmountable problems and has been identified as one of the key factors in the
literacy crisis in Southern Africa. Wagner (2000: 16) and many others report that a significant proportion of students in primary schools are illiterate in their native language(s).
Moreover they do not necessarily succeed in attaining a sufficiently high competence in
the second language either. Instead poor proficiency levels in the second language are
the norm rather than the exception. Given that the second language (L2) is usually the
language of instruction, low proficiency in L2 has been identified as a primary reason
for low throughput and graduation rates at all levels of the educational system (primary,
secondary and tertiary) with profound consequences for employment opportunities.
It is widely acknowledged that, despite decades of literacy campaigns by UNESCO
and others and efforts like Education for All (EFA) (Triebel 2001: 21), attempts to
ensure that what is often referred to as the ‘spread of literacy’ in Africa have failed.
(Bloch 2006: 3)
The current research project attempts to contribute to alleviating the literacy crisis in
Southern Africa by suggesting that the acknowledgment of Southern African ways of
storytelling might lead to the emancipation of African learners on a number of different levels and may contribute to an attitudinal change in speakers of African languages
with respect to their primary language(s). In the outcome, it is anticipated that it might
be possible to redefine the term “quality education” to mean education that integrates
Southern African knowledge systems like Southern African storytelling. Such curricula
reforms could in turn help to achieve quality education in Southern Africa as they might
enable more learners to exit the educational system as economically productive, healthy,
secure and active citizens.
52
4 What about Southern African story grammar?
2 Possible differences between the “canonical” story
grammar and a Southern African Story Grammar
The “western” story grammar tradition suggests that a “canonical” story is comprised
of a setting and an episode system and that stories basically share the same universal
micro- and macro-structural features (Mandler et al. 1980). On a macro-structural level
it is assumed that a “canonical” story encompasses all of the traditional story grammar
elements as identified by Stein & Glenn (1979).1 Table 1 presents these elements.
Table 1: Story grammar elements according to Stein & Glenn (1979)
Story grammar element
Definition
Setting
The spatial and/or temporal location where
story events take place; the introduction of
the main characters, the protagonist, and the
spatio-temporal context.
An event which typically introduces a state-ofaffairs that is ‘out of the ordinary’ for the protagonist, i.e. the occurrence that influences the main
character to action.
An affective or emotive reaction to the initiating
event. Indicates the thought(s), feeling(s) of the protagonist in response to the initiation event; may include an interpretation of the event, formulation of
a goal and serves to motivate action.
A set of intentions formed in the mind of the person affected by the initiating event. Indicates the
intended action of the protagonist (the announcement of the intended action).
The protagonist’s effort to execute the plan. Indicates the overt actions of the protagonist in pursuit
of the goal.
The attainment or non-attainment of the goal, or
other events that are the result of the attempt.
Any emotional or evaluative response by the protagonist to the preceding chain of events.
Initiating event
Protagonist’s internal response
Plan
Attempt
Consequences or outcomes
Resolution or outcome
However, an investigation of relevant literature (e.g. Canonici 1990; Makgamatha 1991;
Motshwari 1998; Obiechina 1992) reveals that Southern African folktales from various
1
These elements are summarized by Griffith et al. (1986: 541).
53
Heike Tappe
language groups differ both in the function and – importantly – terms of their macroand microstructural features from this “universal” schema.
Against the background of these insights and a corpus analysis of popular folktales
from Malawi and South Africa (Tappe et al. in preparation), we propose in Table 2 a first
approximation to a Southern African story grammar. Importantly the elements setting
and internal response of the CSG are demoted in our proposal of a Southern African Story
Grammar, while the elements traditional opening and traditional closing are added.
Table 2: Story grammar elements of a proposed Southern African Story Grammar
Story grammar element
Definition
Traditional opening
The traditional opening initiates the dialogue between
the story teller and the audience and reaffirms the
readiness on both sides for the commencement of the
story telling event.
An event which typically introduces a state-of-affairs
that is ‘out of the ordinary’ for the protagonist, i.e. the
occurrence that influences the main character to action.
A set of intentions formed in the mind of the person
affected by the initiating event. Indicates the intended
action of the protagonist (the announcement of the intended action).
The protagonist’s effort to execute the plan. Indicates
the overt actions of the protagonist in pursuit of the
goal. The description of the execution of the plan typically involves repetitions and dialogues which reflects
the performative character of the storytelling.
The attainment or non-attainment of the goal, or other
events that are the result of the attempt.
Moral lesson(s) that emanate from the story and which
are of general/communal significance.
The story is concluded by a traditional formula which
signals the return to the real world and binds any mystical creatures to the story realm.
Initiating event
Plan
Attempt
Consequences or outcomes
Resolution or outcome
Traditional ending
The setting (i.e. the spatial and/or temporal location where story events take place, the
introduction of the main characters, the protagonist, and the spatio-temporal context)
seems to be demoted or absent in Southern African storytelling which places the emphasis on communal knowledge and universal applicability of the story. The second demoted
element is the internal response of the protagonist to the initiating event: Thought(s) and
feeling(s) are not necessarily ascribed to the protagonist by the story teller. This may be
54
4 What about Southern African story grammar?
a function of the performative aspect of the storytelling where the feelings of the protagonist are enacted rather than verbalised; it may also reflect a taboo.2 Moreover, the
main characters seem to be conceptualised representatives of a particular type of person
(e.g. a young man, a girl), rather than individuals, to allow for maximum identification
with the protagonist. In line with the demotion of the internal response element, the
protagonist’s emotional or evaluative response to the chain of events as it unfolds in
the story also seems to be demoted in Southern African storytelling, whereas the moral
lesson(s) that may be gained from the story events seem(s) to be promoted. Importantly,
the performative character of Southern African storytelling also implies the use of repetitions at various levels of granularity (word, phrase, and paragraph) and the inclusion of
dialogues as a stylistic means in the main section of the story, i.e. the attempt (Canonici
1990; Makgamatha 1991; Motshwari 1998; Obiechina 1992; Tappe et al. in preparation).
3 The relevance of narratives for literacy development
and the consequences of a misalignment between the
“canonical” story grammar and language-specific story
grammars
Bloch (1999: 46) highlights that storytelling – just like reading – exposes children to a
special form of language which is holistic, rich and complex. In a later publication, she
elaborates on this point; “[t]his allows them [the children] to tune into the rhythms and
structures of language and broadens their conceptual worlds and their vocabulary to
express themselves” (Bloch 2006: 11). Moreover, the repetitive and rhyming play with
language that is characteristic for storytelling may aid in developing a child’s phonological awareness, which has been acknowledged as a critical literacy skill by a number of
researchers (e.g. Goswami & Bryant 1990; Duncan et al. 2013).
In addition, Bloch (2006: 13) emphasises the role that narratives play in identity building and in emotional development, which adds further weight to their significance in
child development. Referencing authors like Meek (1992: 44) and Egan (2005: 10); Bloch
points out that through narratives children learn to overcome “egocentric feelings” and
develop a communal sense. This aspect of communal identity is a central aspect of traditional Southern African folktales, which are means to preserve and to express “the
collective memories of a group, amassed over a long period of time” (Ntuli 2011: 65).
However, the questions that arise in the current context are: How can the benefits
of the “holistic, rich and complex” nature of narratives be unlocked to further literacy
development if said narratives are in a foreign language and – importantly – follow a
foreign macro-structure? How can identity building happen if the narratives that children are exposed to and are required to reproduce in an educational setting do not follow
2 In an informal survey, students and colleagues from Lesotho, Malawi, South Africa, Zambia and Zimbabwe
suggested that it may be widely considered inappropriate to verbalise one’s feelings and to ascribe feelings
to others. This observation needs further research.
55
Heike Tappe
a child’s story telling experiences at home? How “naturalistic and ecologically valid” is
the assessment of children’s narratives in Southern African educational systems, which
seem to be deeply rooted in the “western” story grammar tradition and which do not
seem to widely recognise the narrative traditions of the children’s primary languages?
While many researchers emphasise that it is necessary for teachers to recognise the
cultural contexts that learners originate from in order to be able to aid them in appropriate ways (e.g. Heath 1983; Michaels 1991; Bloch 2006; Souto-Manning 2013) and to do
them justice in assessments, this insight does not seem to have widely filtered through
into assessment tool development and curriculum design. On the contrary, research
which addresses the question of how a misalignment between the indigenous narratives
and “mainstream” narratives may affect literacy development in Southern African children is largely outstanding (see, however, e.g. Obiechina 1992 and Bloch 2006).
Research findings, e.g. from the United States of America, clearly indicate a need
to rectify this oversight. Westby et al. (2002) analyse the situation of academically underperforming students from an original people’s background in the USA, where more
than 30% of all students are not Anglo-American, whereas the majority of teachers are
whites of European descent. The authors emphasise that the indigenous narratives that
the children are exposed to at home differ drastically form the narratives they hear in
the media and at school. The task of reconciling the often conflicting demands of the
two narrative traditions is left to the pupils, much to their disadvantage. Westby et al.
(2002: 238) point out that research on the misalignment between narrative practises at
school and narrative practises in the pupils’ primary languages is predominantly available on US-American children from “African-American and Hispanic backgrounds (Gee
1989; Gutierrez-Clellen 1995; Hester 1996; Hyter & Westby 1996; Labov & Waletzky 1967;
Shuman 1986), but very little is available on children from other cultural/linguistic backgrounds.” Even though it seems as if an elaborate academic discourse about the cultural
differences of students’ schooling experiences and the significance of such differences is
readily available, the dominant paradigm appears to be one where an assimilation to the
mainstream is globally more highly prized than acknowledging multi-cultural diversity
(see, e.g. Souto-Manning 2013 for an identical assessment of the situation).
4 The psycho-linguistic “reality” of a Southern African
Story Grammar: Findings from case studies conducted
in Malawi and South Africa
In order to provide evidence for the psycho-linguistic “reality” of a Southern African
story grammar the author embarked on a research project that currently operates in
Malawi and South Africa and is envisioned to bring together researchers and to include
findings from as many sub-Saharan countries and languages as possible. To date our research has yielded a number of surprising results which suggest that despite the prevalence of modern media, urbanisation and the language hegemony of European languages,
Southern African children are still aware of Southern African folktales and that these
56
4 What about Southern African story grammar?
folktales and the underlying story grammar shape the children’s ways of (re-)telling
stories.
The research methodology that we use entails that children in the age range between
10 and 12 answer a language questionnaire to assess their language background and
their experiences with different types of narratives (oral, written, in the form of films).
Thereafter, we present the children either with a wordless picture book or we show them
wordless video clips, which the children narrate first in their one dominant language, and
after a period of about three weeks, in their other dominant language (i.e. currently an
African language and English).3 The children narrate their stories to an interlocutor who
is a speaker of the target language and who (from the children’s perspective) does not
know the stimulus material.
In the first phase of the project we use stimulus material that is not culturally “adequate” for children from Southern Africa which reflects a common experience that they
have in the educational setting; in the second phase we will use culturally adequate material and compare the findings from both phases.
Currently we have data from retellings of two non-verbal video clips4 and four prerecorded audio files which narrate the stories depicted in the video clips in either Chichewa or English. We also elicited narratives by employing the widely used wordless
picture book “Frog where are you” (Mayer 1969). Some of the characters in the selected
stories – land animals (mole, deer, gopher) as well as sea creatures (starfish, crab, seahorse) – and some of the objects used (a rocket) are alien to the children who participate
in our investigation.
4.1 Exposure to indigenous story telling
Our questionnaire studies indicate that children in both Malawi and in South Africa may
still be exposed to African folktales on a regular basis. Tappe & Hara (2013) found that 108
of the 127 Malawian children who participated in their investigation and who live in the
city of Lilongwe in central Malawi heard Malawian folktales on a regular basis. Similarly,
35 out of 41 South African children from urban Durban who are currently participating
in the project rated their exposure to isiZulu folktales on a Likert scale from 1 (I do not
agree) to 5 (I totally agree) by choosing 5; while another three children chose 4. One child
whose primary language is SiSwati chose 3 and two children, whose primary language
is isiXhosa chose 1. In addition many of the children volunteered to tell us folktales and
were very excited when they heard that we are interested in traditional storytelling. This
is in stark contrast to the often heard assumption by parents, teachers and headmasters
that children who grow up in the urban centres do not know and are not interested in
indigenous folktales.
3
The children who participate in our project are often multilingual. Most of them do have two dominant
languages, however, one of which is an African language and the other is the language of teaching and
learning (LOTL). For the Malawian and South African children in our current sample the LOLT is English.
4
“The mole and the rocket” (Miler 1966, accessible at: http://www.veoh.com/watch/v74680117WxTgAWxp)
and “The little mole and the radio” (Miler 1968 accessible at: http://www.veoh.com/watch/
v74680119r3m6t4Rd).
57
Heike Tappe
4.2 The presence of a South African story grammar in the children’s
narratives
Overall, our results indicate that rather than producing stories that adhere to the “western”/ “canonical” story grammar (see Table 1) (Stein & Glenn 1979), the children produce
narratives that reflect features prevalent in Southern African folktales and which are
captured in our first approximation to a Southern African story grammar (Table 2).
The main differences between the two story grammars are reflected in the narratives
that we elicited. As an illustration, Table 3 presents a summary of the findings from
Tappe & Hara (2013) and Hara (2014) with respect to narratives of the Malawian children in our investigation. The most striking differences pertain to the “canonical” story
grammar elements setting and internal response. We found that the children did not produce these in the frequency that one would have expected if the children followed the
“canonical” story schema.
Table 3: The realization of story grammar elements in the Malawian narratives
Element
Opening formula
Setting
Initiating event
Protagonist’s internal response
Plan
Attempt
Consequences or outcomes
Resolution or outcome
Closing formula
Western story African story
grammar
grammar
–
✔
✔
✔
✔
✔
✔
✔
–
✔
–
✔
–
✔
✔
✔
✔
✔
Realised
Chichewa
English
37%
23%
✔
2.49%
✔
✔
✔
✔
33%
39%
17%
✔
5.25%
✔
✔
✔
✔
32%
In stark contrast to these percentages, Acker (2012)5 who investigated South African
monolingual English or Afrikaans speaking children between the ages of five to nineand-a-half years old, found her participants predominantly inserted settings in their
narratives elicited by a wordless picture book (Acker 2012: 80). In particular, 55% of the
five to six year olds, 88% of the six to seven year olds and 100% of the eight-and-a-half
to nine-and-a-half year olds in her sample started their narratives with a setting. Moreover, Acker (2012: 57) found that 50% of her five to six year olds, 65% of her six to seven
year olds, and 68% of her eight-and-a-half to nine-and-a-half year olds described internal
responses of the protagonist.
5
58
There are no comparative data from monolingual Malawian children with a primary language of European
origin available. We investigated in at a number of schools in Lilongwe whether they had any student
enrolment that fits these criteria and received exclusively negative responses.
4 What about Southern African story grammar?
About 30% of the Malawian children inserted an opening and/or a closing formula
as would be expected in traditional Southern African storytelling. The formulae that
the children use are clearly no imitation from the stimulus material. Examples of such
formulae are provided in Table 4.6
Table 4: Opening/closing formulae in the Malawian children’s narratives
Chichewa opening formulae
Chichewa closing formulae
Padangokhala ‘once upon a time’
Nkhani yanga yathera pomwepo, nkhani
yathera pomwepo ‘this is the end of the
story’
Ndamaliza ‘I have finished’
Basi ‘That’s all’ or other similar versions.
Ndi zomwe ndinamvapo ‘That’s how I
heard it’
Kalekale ‘long time ago’
Tsiku lina ‘one day’
Panali or padali or kunali or kudali
‘There was (once)’
In addition to the elements already discussed, the children’s narratives in our corpus
appear to generally diverge from the stories presented to them as stimuli in that their narratives contain a greater number of repetitions and different repetitions than are present
in the stimulus material. Such repetitions occur both at word, phrase and sentence level.
Moreover, the children tend to insert lively dialogues into the stories that are also not
part of the stimulus material. Both dialogues and repetitions are important stylistic elements of traditional Southern African story telling; see e.g. Obiechina (1992: 218) for the
role of repetitions in Southern African folktales.
Our general observation so far is that children both in Malawi and in South Africa
deviate from the stimulus material in that they have a tendency to both omit elements
that were present in the stimulus material and to add elements that were not present in
the stimulus material in a systematic way that reflects features of the proposed Southern
African Story Grammar. We also find that the children who participate in our project
sometimes try to “assimilate” the stimulus material to their background knowledge: In
one narrative, e.g. the protagonist of the “frog where are you” book is placed on a farm
and upon arriving in the forest he meets an “antelope” rather than a deer.
5 The relevance of our findings in an educational setting
Language related obstacles (extremely low literacy rates; insufficiently developed textual
production skills; low levels of discourse/text comprehension skills) have been identified
to be a major hindrance to educational success in the developing world where the “lack
of literacy” appears to be paralysing the developing economies. Indeed national and international efforts to measure literacy in Southern African countries have consistently
6 Compare
Hara 2014: 247 and 250.
59
Heike Tappe
returned devastatingly bad figures over a number of years and across a number of different countries. Smith & Barrett (2011: 23) report on the Progress in International Reading
Literacy Study (PIRLS) 2006 survey, which found that “less than 22% of Grade 4 students
in South Africa […] achieved the “lowest international benchmark” compared to more
than 95% in North American and Western countries (for more detail see Mullis et al. 2007:
69).”
Successful intervention is thus largely overdue. The integration of African storytelling
into Southern African school curricula may be one very powerful intervention. Past and
current attempts to promote Southern African languages at school level seem to have
mainly focussed on terminology development to transform the Southern African languages into languages of academic learning and writing (which – due to budget constraints – has been a very slow process). However, I want to argue that the mere translation of existing teaching materials into the African languages does not solve the problem
that narratives (and maybe other text types, too) may have different macro-structural features in the European and in the African languages. Hence a translated text may still not
alleviate the misalignment between the “canonical” story grammar that is currently required in the educational setting and a Southern African story grammar that is acquired
during pre-school socialisation.
Tragically, however, we do not seem to acknowledge that the Southern African languages have a particular value in the educational setting, which is in part the legacy of
colonial education acts (like the “Bantu education act”7 in South Africa). The outcome
seems to be a “rhetoric of deficiency” with dire consequences for the majority of learners in Southern African countries. They seem to enter school feeling doubly undervalued.
Sentiments like: “My primary language is neither a language of teaching and learning nor
a language of prestige” and “My primary language does not have any value in academia;
it still has to be developed” seem to be prevalent among speakers of African languages.
However the perception that one’s abilities are valuable has been found to be tightly
linked with achievement motivation (e.g. Hardré 2007; Hardré et al. 2007). Achievement
motivation is one of the key ingredients of academic success (e.g. Beck 2004; Hidi &
Harackiewicz 2000; Schunk et al. 1996).
Hence the acknowledgment of existing indigenous narrative language skills could
strengthen the self-worth and the achievement motivation of Southern African children.
Narratives have been identified as means of “self-presentation” because “identity work”
is “being accomplished by the narrating subject” (Baynham 2000: 100). If we were to
strengthen and nourish the narrative skills that Southern African children have in their
native languages, we could amplify their academic skills development in two ways: First,
indirectly, by strengthening their achievement motivation (as explained above) and second, directly, through promoting their narrative skills themselves. Narrative skills have
consistently been found to be closely linked to academic success (O’Neill et al. 2004) and
to reading comprehension development (Cain 2003; Oakhill & Cain 2012). Fostering narrative growth in the early years is hence of utmost important to academic achievement
7 The
major provision of the Bantu Education Act (Act No. 47 of 1953; later renamed the Black Education
Act, 1953) was enforcing racially separated educational facilities.
60
4 What about Southern African story grammar?
and alleviating illiteracy. In conclusion the current paper proposes that bringing stories
that the majority of our children may be able to relate to into the classrooms across
Southern Africa may be a powerful tool to encourage children to engage with language
and with each other across linguistic and cultural divides. Such a reform would help to
alleviate learner frustration and might bolster learner motivation which are essential
prerequisites for successful learning.
Acknowledgements
The researcher is grateful for an ACAL45 travel grant which facilitated her conference
attendance in Lawrence, Kansas. Suggestions and comments from the conference audience and two anonymous reviewers aided enhancing the current contribution. Remaining shortcomings are the author’s sole responsibility. I would also like to thank the children who participate in our project as well as my co-workers, Agness Hara, Mandisa
Ndlovu, Thandeka Maphumulo, Mbali Mntungwa and Megan Govender. Funding from
the South African Research Foundation (NRF) for the funding period 2015–2017 is gratefully acknowledged (CSUR93632: Southern African text and discourse structures and their
relevance for education).
References
Acker, Twanette. 2012. The patterns of development in generated narratives of a group of
typically developing South African children aged 5 to 9 years. Stellenbosch MA thesis.
Anderson, Kimberley & Cay Evans. 1996. The development of the canonical story grammar model and its use in the analysis of beginning reading computer stories. Reading
Improvement 33. 2–15.
Baynham, Mike. 2000. Narrative as evidence in literacy research. Linguistics and Education 11(2). 99–117.
Beck, Robert C. 2004. Motivation: Theories and principles (5th ed.) Upper Saddle River, NJ:
Prentice.
Bloch, Carole. 1999. Literacy in the early years: Teaching and learning in multilingual
early childhood classrooms. International Journal of Early Years Education 7(1). 39–59.
Bloch, Carole. 2006. Theory and strategy of early literacy in contemporary Africa with special reference to South Africa. Oldenburg (Germany): University of Oldenburg. (Doctoral thesis).
Cain, Kate. 2003. Text comprehension and its relation to coherence and cohesion in children’s fictional narratives. British Journal of Developmental Psychology 21. 335–351.
Canonici, Noverino N. 1990. Oral studies in South Africa. Durban. (Master’s thesis).
Duncan, Lynne G., São Luís Castro, Sylvia Defior, Philip H. K. Seymour, Sheila Baillie, Jacqueline Leybaert, Philippe Mousty, Nathalie Genard, Menelaos Sarris, Costas
D. Porpodas, Rannveig Lund, Baldur Sigurðsson, Anna S. Þráinsdóttir, Ana Sucena &
Francisca Serrano. 2013. Phonological development in relation to native language and
61
Heike Tappe
literacy: Variations on a theme in six alphabetic orthographies. Cognition 127. 398–
419.
Egan, Kieran. 2005. An imaginative approach to teaching. San Francisco: Jossey-Bass.
Evans, Rinelle & Ailie Cleghorn. 2014. Parental perceptions: A case study of school choice
amidst language waves. South African Journal of Education 34(2). 1–19.
Gee, James Paul. 1989. Two styles of narrative construction and their linguistic and educational implications. Discourse Processes 12(3). 287–307.
Goswami, Ursula & Peter E. Bryant. 1990. Phonological skills and learning to read. Hillsdale NJ: Laurence Erlbaum Associates.
Griffith, Penny L., Danielle N. Ripich & Sondra L. Dastoli. 1986. Story structure, cohesion, and propositions in story recalls by learning disabled and non-disabled children.
Journal of Psycholinguistic Research 15(6). 539–549.
Gutierrez-Clellen, Vera F. 1995. Narrative development and disorders in Spanishspeaking children: Implications for the bilingual interventionist. In Hortencia Kayser
(ed.), Bilingual speech-language pathology: An Hispanic focus, 97–128. San Diego, CA:
Singular Publishing Group.
Hara, Agness. 2014. Text comprehension in multilingual children: Mental representation
and narrative text structure. Durban (South Africa): University of KwaZulu-Natal. (Unpublished Doctoral Dissertation).
Hardré, Patricia L. 2007. Preventing motivational dropout: A systemic analysis in four
rural high schools. Leadership and Policy in Schools 6(3). 231–265.
Hardré, Patricia L., Michael H. Crowson, Teresa K. Debacker & Deborah White. 2007.
Predicting the academic motivation of rural high school students. The Journal of Experimental Education 75(4). 247–267.
Heath, Shirley Brice. 1983. Ways with words: Language, life and work in communities and
classrooms. Cambridge: Cambridge University Press.
Hester, Eva Jackson. 1996. Narratives of young African American children. In Alan
Kamhi, Karen E. Pollock & Joyce L. Harris (eds.), Communication development and
disorders in African American children, 227–245. Baltimore, MD: Paul H. Brookes Publishing.
Hidi, Suzanne & Judith M. Harackiewicz. 2000. Motivating the academically unmotivated: A critical issue for the 21st century. Review of Educational Research 70. 151–
179.
Hyter, Yvette D. & Carol E. Westby. 1996. Using oral narratives to assess communicative
competence. In Alan Kamhi, Karen E. Pollock & Joyce L. Harris (eds.), Communication
development and disorders in African American children, 247–284. Baltimore, MD: Paul
H. Brookes Publishing.
Labov, William & Joshua Waletzky. 1967. Narrative analysis: Oral versions of personal
experience. In June Helm (ed.), Essays on the verbal and visual arts, 12–44. Seattle:
University of Washington Press.
Makgamatha, Phaka Moffat. 1991. Characteristics of the Northern Sotho folktales: Their
form and structure. Johannesburg & Cape Town: Perskor.
62
4 What about Southern African story grammar?
Mandler, Jean M., Sylvia Scribner, Michael Cole & Marsha De Forest. 1980. Cross-cultural
invariance in story recall. Child Development 51. 19–26.
Mayer, Mercer. 1969. Frog where are you? New York: Dial Press.
Meek, Margaret. 1992. Literacy: Redescribing reading. In Keith Kimberley, Margaret
Meek & Jane Miller (eds.), New readings: Contributions to an understanding of literacy,
224–234. London: A & C Black.
Michaels, Sarah. 1991. Hearing the connections in children’s oral and written discourse.
The Journal of Education 167(1). 36–56.
Miler, Zdenek. 1966. The mole and the rocket. London: Golden Pleasure Books.
Miler, Zdenek. 1968. The mole and the radio.
Motshwari, Julia Jakoentle. 1998. Magic and its significance in Tswana folktales. Johannesburg. (Master’s thesis).
Mullis, Ina V. S., Michael O. Martin, Ann M. Kennedy & Pierre Foy. 2007. PIRLs 2006
international report: IEA’s progress in international reading literacy study in primary
schools in 40 countries. Chestnut Hill, MA: TIMSS & PIRLS International Study Center,
Lynch School of Eduation, Boston College.
Nkosana, Leonard B. M. 2011. Language policy and planning in Botswana. The African
Symposium 11(1). 129–137.
Nomlomo, Sylvia Vuyokazi. 2010. Parents’ choice of the medium of instruction in Science: A case study of one primary school in the Western Cape, South Africa. In Birgit
Brock-Utne, Zubeida Desai, Martha A. S. Qorro & Alan Pitman (eds.), Language of
instruction in Tanzania and South Africa—Highlights from a project, 1–2. Rotterdam:
Sense Publishers.
Ntuli, Cynthia D. 2011. The transmission and reproduction of folktales with special reference to Nanana Bosele. Southern African Journal for Folklore Studies 21(2). 63–75.
Oakhill, Jane V. & Kate Cain. 2012. The precursors of reading ability in young readers:
Evidence from a four-year longitudinal study. Scientific Studies of Reading 16. 91–121.
Obiechina, Emmanuel. 1992. Narrative proverbs in the African novel. Oral Tradition 7(2).
197–230.
O’Neill, Daniela K., Michelle J. Pearce & Jennifer L. Pick. 2004. Preschool children’s narratives and performance on the Peabody Individualized Achievement Test Revised:
Evidence of a relation between early narrative and later mathematical ability. First
Language 24. 149–183.
Orman, Jon. 2008. Language Policy and nation building in post-apartheid South Africa.
Berlin & New York: Springer.
Schunk, Dale H., Judith R. Meece & Paul R. Pintrich. 1996. Motivation in education: Theory,
research, and applications. Englewood Cliffs, NJ: Prentice.
Shuman, Amy. 1986. Storytelling rights: The uses of oral and written texts by urban adolescents. Cambridge: Cambridge University Press.
Smith, Michèle & Angeline M. Barrett. 2011. Capabilities for learning to read: An investigation of social and economic effects for Grade 6 learners in Southern and East Africa.
International Journal of Educational Development 31. 23–36.
63
Heike Tappe
Souto-Manning, Mariana. 2013. Competence as linguistic alignment: Linguistic diversities, affinity groups, and the politics of educational success. Linguistics and Education
24. 305–315.
Stein, Nancy L. & Christine G. Glenn. 1979. An analysis of story comprehension in elementary school children. In Roy Freedle (ed.), New directions in discourse processing,
53–120. Norwood, NJ: Ablex.
Tappe, Heike & Agness Hara. 2013. Language specific narrative text structure elements
in multilingual children. Stellenbosch Papers in Linguistics (SPIL) 42. 297–331.
Tappe, Heike, Mbali Mntungwa & Agness Hara. in preparation. The presence of a Southern
African story grammar in selected folktales from Malawi and South Africa.
Tikly, Leon & Angeline M. Barrett. 2011. Social justice, capabilities and the quality of
education in low income countries. International Journal of Educational Development
27. 3–14.
Triebel, Armin. 2001. The roles of literacy practices in the activities and institutions of
developed and developing countries. In David R. Olsen & Nancy Torrance (eds.), The
making of literate societies, 19–53. Oxford: Blackwell Publishers.
Wagner, Daniel A. 2000. Thematic studies: Literacy and adult education. UNESCO World
Education Forum: Education for All 2000. Paris: UNESCO. http://unesdoc.unesco.org/
images/0012/001233/123333e.pdf, accessed 2014-05-10.
Westby, Carol E., Celia Moore & Rosario Roman. 2002. Reinventing the enemy’s language: Developing narratives in Native American children. Linguistics and Education
13(2). 235–269.
64
Chapter 5
How multilingual policies can fail:
Language politics among Ethiopian
political parties
Mehari Zemelak Worku
Addis Ababa University
Because language has instrumental as well as symbolic values, the issue of language will
always have a political aspect (Smith 2008). Often, the choice of language and its use is construed as one of the central traits to people’s definition of themselves. Besides, any given
state must decide or determine the language that it deems appropriate to carry out its development and to generate, disseminate and enrich the knowledge necessary for such development. However, the case grows problematic when it comes to Sub-Saharan Africa where
“every language carries a distinct and weighty baggage” of identity (Obeng & Purvis 1999).
The decision was not easy for different regimes in Ethiopia, home of more than 80 ethnic
groups (CSA 2008). The three consecutive regimes which have ruled the country for the last
75 years followed different paths in addressing this diversity management question. The
reframing of the country under ethnic federalism, which legislates Amharic as the working language of the federal government (hereafter WL) and guarantees the right of each
ethnic state to decide its own WL, is the recent attempt to respond to the same politics of
recognition. However, dissatisfied voices regarding the current language policy (hereafter
LP) can still be heard among political groups. Some see it as ‘not enough’ while others see
it as Balkanization. Despite a few research efforts and publications on the LPs of the consecutive governments of Ethiopia, there has been no research done on the alternative policies
and options available among the political parties or their relative value as LPs. Thus, the
grand objective of this study is to survey, analyze and evaluate the linguistic proposals of
Ethiopian political parties in government, education, and endangered languages.
Mehari Zemelak Worku. How multilingual policies can fail: Language politics among Ethiopian
political parties. In Jason Kandybowicz, Travis Major, Harold Torrence & Philip T. Duncan (eds.),
African linguistics on the prairie: Selected papers from the 45th Annual Conference on African
Linguistics, 65–83. Berlin: Language Science Press. DOI:10.5281/zenodo.1251716
Mehari Zemelak Worku
1 Language politics in Ethiopia: Historical overview
In this section I will briefly discuss the issue of language throughout Ethiopian political
history.1 A compact version of the history will be presented with a focus on three sections: pre-1974 (before the outbreak of the student-led Ethiopian Revolution), 1974–1991
(the time where the military junta called the “därg” stayed in power) and post-1991, the
era of the incumbent EPRDF.
1.1 Pre-1974: Language in imperial Ethiopia
“In the historical, political and literary fields, two languages have dominated Ethiopian
Studies: Amharic and Geez” (Bender 1976). Along with their dominant status in the field
of Ethiopian Studies, Bender’s words can be extended to show how overriding the two
languages were in the overall social life of Ethiopia. As the antiquities from the Axumite
Dynasty indicate, Geez was the official indigenous language of the Axumite Dynasty
(Philipson 2012). This official domain had been kept to Geez until Amharic took over
as lәsanä nәgus ‘the language of the king’ some time during the reign of the Zagwe
Dynasty in the 11th–13th centuries (Awgchew 2009). We do not have any document to
answer why and how Geez took over the higher domains as the official language in that
time. Nevertheless, we can say it can be one of the two cogent reasons given by Cooper
(1976): as a result of government-implemented official LP or as a consequence of societal
dynamics among speakers of different languages, or maybe both.
Irrespective of efforts by evangelists to reduce other Ethiopian languages into writing
like Onesimos Nesib, an Oromo missionary who translated the Holy Bible into Affan
Oromo2 using the Ethiopic script in the early 20th century, Amharic and Geez continued to confluence dominantly in the educational, religious and political domains of the
region until the 20th century, when western education stepped in to the country with
the complete support of emperor Menilik II (ኃይለየሱስ 2008). The policy of the Ethiopian
kingdom regarding other languages was only to be found in the indigenous traditional
collection of religious and secular laws known as the Fәtha Nägäst (law of kings). The
code has been used in the country for several centuries and contains legal provisions
for speakers of other indigenous languages to have judges speaking their native tongues
(Tzadua 1968 cited in Cooper 1976).
The boost of modern education and its legacy during the reign of emperor Haile Selassie I created a new domain for Amharic and Geez. Meanwhile other indigenous languages, regardless of their powers in numerical terms, were kept only to non-official
domains. The education arena, in its early stages, was generally a field for the learning
1
In this regard the works of Cooper (1976); McNab (1990); Bulcha (1998); Getachew & Ado (2006); Heugh et
al. (2007); Smith (2008); Zemelak (2011) and Leyew (2012) are worth reading for a more detailed discussion
on the typology of the LPs of different regimes in Ethiopia. Even though it is not going to be confined to
their outlooks, the timing frame of the discussion in this paper is basically sketched out by the marked
works.
2
Cobarrubias (1983) wrongly writes “Gallah [sic] in Ethiopia” (Galah is a derogatory reference to mean
Oromo) as an example for a language “without a writing system.” But, by the time Cobarrubias’ article was
published, Afan Oromo speakers had been using the Afan Oromo Bible for about 90 years.
66
5 How multilingual policies can fail
of foreign languages. Remarkably, it was a stage of competition between French and English, as the envoys of both neighboring colonial powers vied for swift influence over
the Ethiopian crown. Initially, the French were more successful than their British contenders in retaining a positive attitude of the oligarchy towards French. However, this
lasted briefly, only up to 1935 when Fascist Italy declared war on the Ethiopian Empire
and invaded to avenge the humiliating defeat it suffered at the battle of Adowa in 1896.
Bowen & Horn (1976: 610) note that during the Italian occupation, which lasted until 1941,
the burgeoning educational efforts of Ethiopia were interrupted and most schools were
closed. The mission of the few functioning schools was altered from training indigenous
children to be citizens defending the country’s interest in all possible domains, to educating the colonial masters’ Italian children to keep them in touch with their European
“mother civilization.” This offered them a corner of the earth where they could enjoy ‘sun
bathing’, to use Mussolini’s terms. Ethiopians were obliged to stay confined to primary
level education.
With the implementation of the age-old divide et impera policy of oppressors, the
Italians divided the country into six regions according to major ethnic lines. They also
ordered, without any deliberate effort to capacitate the languages in corpus planning,
each of the major indigenous languages to be used in native schools in the respective regions. In addition, the Governor- General of each province had the authority to establish
any other language as additional instructional medium in the schools (Pankhurst 1976:
322, cited in Bowen & Horn 1976; Zewde 2002).
Table 1: Administrative regions and languages of instruction during the Italian
occupation (1935–1941).
* This term is used by the Italians to refer to the Oromo people. However, the
Oromo people do not refer themselves with this word and is considered offensive.
No.
Regions named by the Italians
Languages proclaimed to be media of
instruction
1
2
3
4
5
6
Eritrea and Tigray
Amhara
Addis Ababa
Harar
Galla* and Sidama
Somalia
Tigrinya and Arabic
Amharic
Amharic and Oromifa
Harari and Oromifa
Oromifa and Kafficho
Somali
This LP, nevertheless, was not put into full effect except in Eritrea, which had been a
colony of Italy for about fifty years after the battle of Adowa in 1896. First, the policy was
understood by the natives as a separatist movement to dissuade the country’s noble men
and gentry in to ethnic division to enervate the patriotic movements. Second, during the
Italian occupation the education system was barely functioning. Third, as a result of the
multitude of patriotic movements against them, the Italian Fascists did not have enough
67
Mehari Zemelak Worku
time to establish a lasting educational system. Nonetheless, this time is remarkable because of the deliberate abandonment of any effort in the education sector to promote a
common indigenous language of communication among the variegated linguistic groups
of Ethiopia. The LP of the occupiers came to an end when the Italians were ousted by
the coalition of Ethiopian patriotic forces and the British troops from eastern Africa in
1941.
After the Italians, the British took their shot at attempting to influence the crown.
Since then ‘the British way’ of doing things came to be accepted as the better way in
alignment with His Imperial Majesty Haile Selassie’s I aspirations for a strong, modernized nation. This was particularly reflected when gradually English took away the
education domain that French occupied.3
Amharic had also reclaimed its former dominance over all indigenous languages as
Haile Selassie I aspired to build a unified modern empire under one crown and one dominant language. In the revised constitution of Ethiopia it was stated explicitly that “the
official language of the empire is Amharic.” (Revised Constitution of Ethiopia: 125). It
was a must that new laws be published in Amharic and English in the official gazette of
Ethiopia. The statutes of domestic companies were required to be filled in Amharic or
English. Foreigners were also required to write and speak Amharic ‘perfectly’, in order
to attain Ethiopian citizenship. According to the regulation decreed in 1944 concerning
missionary activities in Ethiopia, missionaries also were obliged to learn Amharic to
pursue their mission for it was outlawed to use other indigenous languages except at
the early stages of the missionary activity. The credo insists on both the missionary and
the pupil to have a working knowledge of the “official language of the empire” (Cited
in Cooper 1976). Such proclamations, charters and codes vividly display the dominant
status Amharic had in the state.
In the 1960s, as a result of external and internal political pressure exacerbating through
time, the state-owned broadcast media had shown some signs of a more relaxed stance
towards embracing indigenous languages other than Amharic. The languages that were
introduced to the government-owned broadcasting radio stations were Afar, Somali,
Tigrinya and Tigre (Smith 2008). However, the time allotment for them was extremely
scanty when compared with the time allotment to Amharic. The imperial administration did not have the political will to abandon the counterproductive effort of unifying
a multilingual country under one language and a unitary state.
3 According to Bowen & Horn (1976) the erosion, nevertheless, began in the 1930s before the Italians stepped
in to the Ethiopian soil. Aleme Eshete (cited in Bowen & Horn 1976) points at one person’s influence on
the crown, Dr. Workneh Eshete, who was taken to India, a British colony by then, and studied medicine.
As Aleme remarks, because of this man’s beliefs in the British system and conviction to make the country
as strong as the Great Britain, a group of 18 Indians were brought to Ethiopia in the 1930 to serve as
technicians, doctors, teachers and other badly needed specialists. Since Indians were not well versed in
French, if not oblivious at all, it was necessary for the students to become literate in English to exploit the
available expatriate resource. However, a few exceptions can be mentioned in relation to the use of local
languages in the Ethiopian primary education. For example, in Swedish and American missionary schools,
mother tongues were used before European languages at the primary level (McNab 1988, cited in Heugh
et al. 2007: 45).
68
5 How multilingual policies can fail
1.2 1974–1991: The Ethiopian Student Movement and the därg on the
language issue
Amharic as an established state language was challenged with the increase in influence of
modern education in the country’s social, economic, and political domains. University
students who were becoming increasingly political espoused strong criticisms against
the regime. The imperial regime did not have the administrative ability among its agents
to undertake any successful amendment to appease the public demand (Zewde 2002).
Left wing Marxism became the dominant ideology among the students. Immediate
recognition of “the oppressed identities of nations and nationalities” 4 was one of the
main demands of the students. The issue reached a boiling point when the Lenin/StalinFanon inspired Wallelign Mekonen published an article entitled “On the Question of Nationalities” on 17th November, 1969 in the Haile Selassie I University student publication,
Struggle. Walelign merged the Leninist/Stalinist idea of the nation with Frantz Fanon’s
justification of violence against colonial oppression. His argument was that Ethiopia was
not a nation; rather it was a collection of nations (Gurage, Somali, Oromo, etc.) that had
been subdued by the “Amhara-Tigre (two of the dominant ethnic groups) hegemony.”
These oppressed nations, he explained, can only reclaim their freedom through “revolutionary armed struggle” (Mekonen 1969). 5 Since then, the language issue has clung to
Ethiopian politics as a symbol of identity recognition for ethno-linguistic groups in the
country.
The imperial regime was toppled in 1974 by a revolution rooted in the student movement, and abetted by the distraught military personnel of the regime. In the wake of the
victorious revolution, a sense of euphoria spread with the prevailing idea of creating a
modernized and prosperous Ethiopia, where all people, irrespective of their linguistic
identity, would be treated equally. The trajectory of ethnic equality spilled over to language, as every linguistic group in the country was to be recognized as equals (Zewde
2002; Gudina 2006; Balsvik 2007).
Consequently, the military council, 6 which hijacked the revolution, 7 waged a literacy
campaign across the country using 16 indigenous languages as media of instruction for
basic literacy. This was in direct contrast to the former overtly monolingual education
policy. The 1976 political program of the National Democratic Revolution of Ethiopia
overtly proclaimed the right of nationalities as the following:
4
It is at this particular time where ethnic groups of the country started to be referred as “nations and nationalities” in the political discourse of the country.
5
For Stalin (1913) one of the defining characteristics of a nation is speaking one common language. After
defining the nation as “a historically constituted, stable community of people, formed on the basis of a
common language, territory, economic life, and psychological make-up manifested in a common culture,”
he asserts that “it is sufficient for a single one of these characteristics to be lacking and the nation ceases to
be a nation.” Wallelign’s denial of the existence of the “Ethiopian nation” as one nation seems to be based
on influence from his Marxist tendency.
6
Often called “the därg,” a word derived from Geez to mean ‘committee’ or ‘council’.
7 The aim of the campaign had two aims: to show the military council’s intention to the linguistic equality
and empowerment of all Ethiopians, to weaken the voice of the students, who were calling for the establishment of civilian administration, by dispersing them in the rural parts of the country as teachers in the
literacy campaign (Balsvik 2007).
69
Mehari Zemelak Worku
The right of self-determination of all nationalities will be recognized and fully
respected. No nationality will dominate another one since the history, culture,
language, and religion of each nationality will have equal recognition in accordance with the spirit of socialism…. Given Ethiopia’s existing situation, the problem of nationalities can be resolved if each nationality is accorded full rights of
self-government. This means that each nationality will have regional autonomy
to decide on matters concerning its internal affairs. Within its environs, it has the
right to determine the contents of its political, economic, and social life use its own
languages and elect its own leaders and administrators to head its own internal organs (Ethiopian Government Programme (1976) cited in McNab 1990).
Despite the overflowing rhetoric, the centralist ideology of the military regime and
its ambition to create a unified Ethiopia was uncompromising to the idea of transferring
power to a civilian government. It did not take much longer for the politics of recognition
and autonomy to ascend to the dome of political discourse as an unanswered question.
Because of its extremely repressive politics the military junta forced most of the leftwing
socialists and ethnic liberation groups, some of whom were already guerilla fighters, to
armed struggle. Later, the military council unsuccessfully tried to transform itself into
a civilian government. In 1987 it promulgated a new constitution, which renamed the
country as the Peoples’ Democratic Republic of Ethiopia. All indigenous languages were
guaranteed equal recognition as languages of the country. Amharic lost its status as the
national language but maintained its degree of dominance as the constitution declared
it as the official Working Language (hereafter WL) of the government.
Under this constitution the country was reframed under 14 autonomous provinces
as way to appease the guerilla fighting ethnic rebel groups. It was too late! The rebels
disparaged the transformation attempt as a theatre to cover all the cantankerous and
assimilationist natures of the dictator Mengistu Hailemariam to pen in the rebels, but
they did not give in.
1.3 Post-1991: Language in the country of “Nations, Nationalities, and
Peoples”
The military regime’s LP was essentially a de jure claim of equal status of all languages
in the country, while the de facto LP was merely a preservation of the imperial LP which
prescribed Amharic for all educational and official domains at the expense of other indigenous languages. Research done by the Ministry of Education in 1986, proves the
language policy of the military regime was merely the extension of the imperial policy, jeopardized the enrolment and efficacy of students whose mother tongue was not
Amharic (cited in Heugh et al. 2007.
On May 28th 1991 the coalition of ethnic cored armed groups under the umbrella of
EPRDF took control of the capital Addis Abäba, ousting the military regime. The questions of autonomy and recognition of linguistic identity were reinvigorated. In July, 1991
a conference was held for the plethora of newly formed as well as old, but highly enervated, political parties. At the conference a charter, which served as a foundation for the
70
5 How multilingual policies can fail
forthcoming constitution, was provided. That was the stance where the political arena,
which used to be entirely occupied by a military junta, officially proclaimed to embrace
several parties to give sufficient significance for all the voices of cultural and political
groups (Zewde 2008; Berhanu 2003). The question of language appeared to be a subject
of notability to the political agendas held by most of the political groups. The historian
Bahru Zewde recalls an event from the conference:
The first order of business in the course of the eventful deliberations of that conference was the determination of the working language. In view of the difficulty
getting interpreters for the multiplicity of languages represented by the ethnonationalist organizations attending the conference, it was decided to adopt Amharic and English. However, organizations that brought their own interpreters could
use their language. The EPLF leader, Issayas Afeworki, took the latter option and
addressed the conference in his native Tegreñña. But what is of particular interest
to this investigation is the ire that he visibly demonstrated at what he felt was a less
than adequate Amharic rendering of his delivery, showing that his Amharic was as
good as, if no better than, his designated interpreter. The spokesman of OLF, also
understandably, addressed the conference in his native Oromo language. But it was
translated not in to Amharic but into English, there by leaving the overwhelming
majority of the national audience in the dark (Zewde 2008: 77).
This is a self-explanatory situation to portray the linguistic sensitivity of the post 1991
political system. In 1992 the new Transitional Government of Ethiopia first proclaimed
the rights of every ethnic group of Ethiopia to use and develop its languages and culture.
This was further strengthened and confirmed in the 1995 Ethiopian Constitution.
In contrast to the 1987 constitution which starts with the phrase “We, the working
people of Ethiopia”, the preamble of the 1991 constitution begins with “We, the Nations,
Nationalities and Peoples of Ethiopia.” 8 The constitution also divides the country into
nine autonomous ethnic states each with its own constitution, flag and the right to selfdetermination including and up to secession (Article 39). This took nobody by surprise
since most of the political players who had a greater role in writing the constitution were
members of the student movement. In the retention of Amharic as the WL of the government and the egalitarian approach to all languages in the country, the new constitution
follows in the footsteps of the 1987 constitution (Article 5: 2). Audaciously supporting
the ethnic cause, the new constitution gives member ethnic states of the federation unsurpassable right to determine their respective WL by law (Article5: 3). Subsequently,
indigenous languages became widely used in primary education, media, administrative
and judiciary systems. In 1994 the Ministry of Education of Ethiopia proclaimed a new
8 However,
the constitution doesn’t give a single hint, let alone definition, to state the difference between
these three terms is. Rather, it defines all the three terms as follows: ”Nation, Nationality or People for
the purpose of this Constitution, is a group of people who have or share large measure of a common
culture or similar customs, mutual intelligibility of language, belief in a common or related identities, a
common psychological make-up, and who inhabit an identifiable, predominantly contiguous territory”
(Article 39: 4). Hence, in the current political discourse of Ethiopia any linguistic group is considered as a
“nation/nationality” (Vaughan 2006).
71
Mehari Zemelak Worku
Education and Training Policy, which declares: “Cognizant of the pedagogical advantage
of the child learning in mother tongue and the rights of nationalities to promote the use
of their languages, primary education will be given in nationality languages” (FDRE
1994). The existing language in education policy frames the entire system as indicated
by the following table.
Table 2: The Existing Language in Education Framework.
No.
Languages
Level of education
Function
1
Mother tongues/
Nationality languages
Primary
Medium of instruction
2
Amharic
Primary up to
secondary level
education
As a subject and
language of country
wide communication
3
English
From grade 1
As a subject
4
English
Secondary and higher
education
As a medium of
instruction
5
One more nationality
and foreign language
All levels
As subjects for
intercultural and
international
communication
6
Mother tongues/
Nationality languages
Teachers training for
primary level education
Medium of Instruction
Currently, there are twenty five indigenous languages used as media of instruction.
Most regions selected indigenous languages spoken by the ‘majority’ of the denizens of
that particular area. It is about twenty years now since this LP favoring indigenous languages has been implemented, and there is a sizeable amount of research reckoning the
effects of this policy. However, it is still considered by the EPRDF government as one
of its achievements from both the linguistic rights as well as pedagogical point of view.
There are political groups who consider the current system as ‘The Answer’ for all questions on linguistic rights. On the other hand, there are others who criticize the system as
Balkanization and enhancing deadly ethnic conflicts. These voices have representations
in the political discourse of the country through political parties which suggest options
through their political programs and manifestos. In the next part of this paper I will try
to expound what policy proposals are forwarded by these political groups of the country.
72
5 How multilingual policies can fail
2 Sampling methodology and conceptual tools
Scarcity of resources would not allow this study to include all the National Electoral
Board of Ethiopia recognized (hereafter NEBE) 79 political parties, since most of them are
regional parties (Board 2009). Particularly the number of regional parties, 63, and their
scattered location all over the country is unmanageable to be taken as a whole. Hence,
this study is done by taking representative samples. According to the NEBE, there are
79 National9 and Regional Political Parties. Out of these 19 are categorized by the NEBE
as National Political Parties while the remaining 60 are labeled as regional. Out of the 19
parties labeled as national 10 were selected randomly.10 The same was done to the parties
categorized as regional. Fourteen regional parties were selected randomly.11 This enabled
the sample to include about 30.4 % of the entire party population. The size of the sample
was intentionally expanded to include parties from different ethnic and ideological bases.
Moreover, it is worth mentioning that the distribution of regional parties is extremely
uneven throughout the regional states of the country. Among the regional parties there
were those which reported to NEBE not to be functioning anymore and thus excluded
from the sampling.
The conceptual tools used in this descriptive study to analyze the types of the LPs as
well as the motivations and ideologies beneath the LPs are garnered eclectically from
Fishman (1972); Cobarrubias (1983); Schiffman (1996) and Patten (2001).
3 Language politics among the functioning political
parties of Ethiopia
In this section an overall assessment of the LPs of the political parties is presented first.
Then, I will deal with the proposals of the political parties about the language that should
take the status of the WL of the federal government. The next focus of the analysis is the
proposals about language in education. Through the analysis besides providing typology
of the LPs motivations and ideologies of the proposals are inferred.
3.1 Overall assessment of the political parties’ LPs
Although the need for a thoroughly thought-out and well planned LP for a multilingual
country such as Ethiopia is unquestionable, out of the 24 political parties included in
this study only seven12 have put relatively clear and implementation orientated LPs. It is
only these seven parties which explicitly propose to legislate, if they are voted to power,
what the federal WL should be or how it should be selected, and what the language
use in education shall be. Moreover, they also suggest establishment and expansion of
9 According
to the Board (2009), national political parties are those whose campaigns are not limited to a
certain ethnic state. In contrast, parties which run only in a certain ethnic state are called regional parties.
10 GSPP, EDP, EPRDF, Forum, EJDFF, AEUP, EPAP, ERaeiP, UEDF, AEDP.
11
ONC, OPDU, DDQPDU, DDDP, OLP, WPDF, HNL, GPRDF, KPC, ANDO, ARDUF, BMPDO, BGDUF, WSDP.
12 EDP, AEDP, EPRDF, OLP, Forum, GPRDF and ONC.
73
Mehari Zemelak Worku
language research centers and the deployment of resources by the federal government to
the corpus development of the languages. Out of the remaining 17 parties 16 of them do
not give detailed and implementation oriented LPs. The remaining one party13 proposes
nothing about the language issue.
Depending on the broadness or narrowness of their LPs, the 16 parties without detailed LPs can also be divided into two sub-categories. First,14 there are ten parties which
present a too-broad and highly generalized claim to ensure the recognition of the languages and cultures, history and identities ethnic groups. Even though these parties
have not offered any detailed implementation frame work, they have displayed support
to mother tongue education at primary level. The second group holds five15 parties, all of
them regional parties, which provided too-narrow LPs in their political programs such
as, language X shall be included in schools as a subject. The proposals in this group neither tell what the medium of instruction at the schools shall be nor what is to be done in
the development of the languages they suggest to be included in the education system;
nor do they say anything about the linguistic rights of other linguistic groups in the area
where these parties aim to gain legitimacy.
3.2 Issue of the federal WL
Out of the seven parties that give relatively detailed and implementation orientated LPs,
two16 propose the promotion of Afan Oromo, to have the WL status beside Amharic.
The reason for such legislation, the two parties forward, is the numerical muscle Afan
Oromo have.17 Three18 parties from this group propose the adoption of an additional WL
beside Amharic without mentioning a particular language. One19 of these three parties
forwarded that the additional WL should be chosen from the indigenous languages based
on its numerical muscle; while the other two20 preferred the selection to be on the basis
of “facilitating science, technology and market exchange.” On the other hand, two21 of
these seven parties stand in peculiarity by not mentioning the promotion of any specific
language overtly; rather they propose egalitarian preservation, usage and development
to all languages in the country. All22 the proposals from the parties grouped under the
non- detailed LPs have the same stand. All propose the preservation of ‘all languages’ in
the country.
As described above, the LPs from these two parties are overt and dedicated to the
promotion of Afan Oromo to be the federal state’s WL. As the language they propose is
13
DDPDO.
HNL, AEUP, GSPP, WPDF, DDQPDO, UEDF, EJDFF, OPDU, EPAP, ERaeiP.
15
ARDUF, WSDP, ANDO, BMPDO, KPC.
16 OLP and ONC.
17 Oromo is the largest ethnic group in Ethiopia (CSA 2008).
18 EDP, AEDP, and FORUM.
19
FORUM.
20 EDP and AEDP.
21
EPRDF and GPRDF.
22 DDQPDO, WSDP, ANDO, HNL, WPDF, BGDUF, BMPDO, OPDU, ARDUF, EPAP, ERaeiP, AEUP, UEDF,
EJDFF, GSPP, KPC.
14
74
5 How multilingual policies can fail
an indigenous one, they are endoglossic LPs. Besides, these overt and promotion LPs of
the parties can be said to follow the Language Maintenance model to formulate their LPs
for the following reasons. First, the aspiration is to preserve and promote the language
and the identity represented by the language through making the language the WL of the
FDRE. This provides a better opportunity to the language and speakers of the language to
preserve their language and identity as something worth preserving as it opens access
to higher domains of state business and economic benefit. In other words, the parties
aspire to win symbolic affirmation and identity preservation simultaneously. However,
because none of the policies proposed by the political parties have explicitly included
anything about the economic, social, cultural and/or political benefits to be gained by
making the language they propose the WL of FDRE, the motivation can be argued to be
sentiment and authentication. The only points the LPs mention to justify the need for
Afan Oromo to be the national language is the numerical muscle the language has and as
part of the “emancipation” of Oromo identity from the “cultural oppression” perceived
by the parties.
The second group is constituted from three parties23 which follow linguistic rationalization model and overtly propose the adoption of an additional WL besides Amharic.
One of the parties in this group proposes the numerical muscle of the language to be the
rationale for selection. The remaining two24 parties, forward that the selection should
be based on “the people’s will” and suitability for science, technology and market exchange. As all the parties in this group favor the selection and promotion to be from
the languages of the nationalities of the country their LPs can be labeled as endoglossic
and promoting. On the other hand, from the point of view of the model provided by
Patten (2001), the policies of these political parties always confirm that all the languages
and identities of nationalities shall be respected and allowed to flourish. With this their
proposals show a major feature of the official multilingualism model.
The third group holds two25 of the detailed and all of the non- detailed parties. These
parties propose nothing as a precise answer to the question of the WL of the federal
government. Instead, they propose the equal preservation and cultivation of languages
and linguistic identities.
Because the parties in this group have no explicitly stated language proposal about the
WL of the federal government, it will be ambiguous to point their sway exactly whether
they have agreed with the status quo or not. This makes their LPs regarding the WL of
the federal government to be a covert one. However, it is also worth mentioning that
they have an overtly stated proposal for an egalitarian preservation of all languages in
the country. Besides, since their policies affirm the preservation and cultivation of the
languages of the “nations, nationalities and peoples” of Ethiopia, it can be said that as a
trajectory they lean towards the indigenous languages. Although all the parties in this
group do not explicitly promote any particular language, their proposals for the dedication of the state towards full support to the development and preservation of the
23 EDP,
AEDP and FORUM.
AEDP and EDP.
25 EPRDF and GPRDF.
24
75
Mehari Zemelak Worku
indigenous languages makes them holders of the promoting and official multilingualism
models. Yet, there are two26 parties in this group that presented the value of recognition of the indigenous languages from an angle of political participation and democratic
rights as follows:
Unless the people are using their own language they would not be able to sufficiently participate in the political system through interpreters. Unless citizens use
their language to express about their culture and history it is impossible to them to
exercise their right to freedom of expression (Press and Audio Visual Directorate/
EPRDF 2002).
The motivations behind the LP proposals of this group can be said to be both sentimental and instrumental. Those who propose the preservation and recognition of the
languages for the sake of identity preservation are the ones with a sentimental motivation and those that propose the recognition and preservation of the languages to be used
by the respective people emphasize the political benefits to be reaped from using one’s
own language as an instrument for communication.
The fourth group is constituted from one regional party27 with no language proposal
at all. This party has suggested no proposal regarding the WL of the federal government.
Yet, it is still possible to argue that the party hasn’t mentioned the language issue in its
political programs may be because, it has a covert LP that is comfortable with the current
LP of the country.
3.3 Language in education
From the parties that presented relatively expatiated language proposals,28 two parties29
explicitly propose one particular language to be used as medium of instruction. These
parties overtly state Afan Oromo to be used as medium of instruction in schools. However, no instrumental benefit that is to be gained from using the language as medium of
instruction in this multilingual country is pointed out by the parties. It can be argued that
such LPs covertly marginalize other languages from being used as instructional media
within the Oromia ethnic state itself since other ethnic groups also live there. In the proposals from these parties there is no mentioned space in education to other languages of
the country. Since the LP is aimed at promoting the language and the linguistic identity
the ideology beneath can be supposed to be vernacularization. The motivation behind
can be, thus, inferred as authentication of the Oromo identity. However, one can contend
that there is an assimilationist and hegemonic tendency in this policy because it tries to
impose a particular language over “all schools” in the vicinity to be ruled by the party.
Such a policy does not seem to look how imposing one’s language in a multilingual arena
would certainly contradict with the economic development and political participation of
26 EPRDF
and WPDF.
DDPDO.
28 FORUM, GPRDF, EPRDF, EDP, AEDP, ONC and OLP.
29
ONC and OLP.
27
76
5 How multilingual policies can fail
minority groups. This in turn can result resentment from other ethno-linguistic groups.
At the worst case scenario, it could invite a total discordance in a region.
From the non- detailed group, two parties30 in the too-narrow sub-group share common feature that they mention a language to be used in education. For instance, the social program of the ARDUF suggests that “Afar language shall be included in the school
curriculum as a subject.” The social program from the KPC also suggests that “Kambata
language shall be introduced and taught.” In the proposals from these parties, there is
nothing stated about the other indigenous languages in the region. In addition, the proposals do not seem to push beyond making the languages they mention to be included
in the school curriculum as a subject. They also do not justify their suggestions based
on the benefits that the policies will bring to the people. Nor they give any additional
policy about developing the languages to sustain their use in education.
The remaining five31 parties in the detailed group, one32 of the parties from the toonarrow and all the parties in the too-broad subgroups accede with one axiom: mother
tongue education at the primary level. Particularly at the primary level education all of
them consent that the medium of instruction shall be the mother tongues of the students.
None of the policies of these parties restrict any indigenous language from being used as
medium of instruction. However, it is worth mentioning that two33 parties in this group
propose to take special consideration towards metropolitan cities in the implementation
of mother tongue education. With such policies the two parties have shown special consideration and a peculiar treatment of the metropolitan cities of the country that are
practically variegated in ethnic composition. This gives space to the interest of people
from different ethnic groups as their children probably pick the ethnic language of their
parents as mother tongue. There is also the possibility for the children to assume the
dominant language of the city as their native language. Besides, in addition to promoting the indigenous languages to be used as media of instruction, such policies also show
caution from imposition of one language over another linguistic community without the
interest and consent of the latter.
On the other hand, from the parties that offered relatively detailed LP, three34 of them
which are among parties that proposed the selection of additional language(s) to the
status of the federal WL, affirm that, after selection of the additional language the selected language(s) of the federal government shall be taught as (a) subject(s) in the entire schools of the country; and shall also serve as media of instruction depending on the
need. They also have conferred the responsibility onto the shoulders of the government
to implement the teaching of the selected language(s) in the regular education system.
Unification is the major motive at play in such LPs because teaching the federal WL in
all schools of the regional states is basically to facilitate communication and social mobility among different language speakers of the federation as one political society. Since
these LPs confer the authority to the regional states to select their respective WLs and
30 KPC
and ARDUF.
EDP, EPRDF, AEDP, FORUM, and GPRDF.
32 ANDO.
33
EDP and AEDP.
34 EDP, AEDP, and FORUM.
31
77
Mehari Zemelak Worku
use them as media of instruction in schools located in the region, there will be no space
for the oppressive assimilationist ideology.
Except for one35 party none of the parties included in this study have proposed an LP
regarding the language to be used as a medium of instruction beyond the primary level
education. On the other hand, except for the above pointed one party and the parties
that propose the maintenance of the selected additional WL of the federal government,
none of the political parties have articulated anything about the medium of instruction
after the primary level education or on second or foreign language choice. It might be
conjectured that they are either comfortable with the status quo, i.e. using English as
medium of instruction, or that they have no alternative policy to offer the public in this
regard.
4 Conclusions and recommendations
Although there exists no political party that overtly proposes the adoption of a monolingual LP, none of the parties can take pride in having a thoughtful, all-encompassing
and strategically implementable LP. These policies have failed in addressing burning linguistic issues in Ethiopia.
The failure of the LPs to address the issue of language beyond primary language education is one instance. Notwithstanding the necessity of language of wider communication (LWC), the best communicative medium in education is the best known language
by both the learners and the teacher: in most of the cases the mothertongue. However,
in Ethiopia, English -a language barely known by the overwhelming majority- takes the
secondary and tertiary level education dominantly as the only official medium of instruction. The reality in Ethiopian schools as well as researches investigating the effectiveness
of this policy, nevertheless, indicate that English, a foreign tongue for both the teachers
and the students, has become the “medium of obstruction” in classes (Stoddart 1986 cited
in Bogale 2009). It is ironic to see none of the political parties addressing this very critical
issue! The other point where all the political parties have failed to address is the issue
of language endangerment in the country. According to UNESCO (2015), 28 Ethiopian
languages are endangered. The numerical muscle of most of Ethiopian languages is in
hundreds and a few thousands. I argue that none of the current multilingual LPs of the
political parties discussed above will benefit these languages. Putting an LP which treats
all languages as equals can by no means be beneficial to these endangered languages,
no matter how endoglossic it is. Equality should not mean treating the advantaged and
the disadvantaged in the same way. Rather, it is supporting the disadvantaged without diminishing the advantaged unfairly. LPs cannot benefit their respective societies
just because they appear to be proponents of multilingualism. Without well thought
and strategically well designed LPs neither empowerment of indigenous languages nor
preservation of endangered languages can happen.
35 OLP.
78
5 How multilingual policies can fail
Since Ethiopia is a multilingual state, with all the centuries old squabbling between
elites of different ethno-linguistic groups, any societal development plan should always
consider language as an issue to be thought thoroughly in planning economic development as well as regional tranquility. Hence, all the political parties should try to work in
cooperation with the concerned academia for a well detailed discussion of the pros and
cons of LPs.36 In addition, beyond the promotion of ethnic identities, parties should look
for policies that will have a positive effect on durable regional stability, since imposing
one’s language over the other will inevitably ostracize the latter from easily accessing
economic development and political participation.
Furthermore, as there is a sizeable number of parties that are calling for an additional
WL of the federal government, all of the people in the political business and stake holders
of the language issue, such as researchers and consultants of policy makers, should begin
to deeply look at the economics of such LP from the perspective of language economics
(Grin 1994; Chiswick 2008). Any proposal regarding the WL of the federal government
should not be a result of incessant lobbying from the political entrepreneurs. A thorough
cost benefit analysis that carefully considers the ethnographic and economic realities of
the country is the only best way to design an effective LP.
Acknowledgments
I would like to say the age old adage “thank you” for the ACAL45 conference organizers
at KU, particularly Dr. Jason Kandybowicz, for the travel grant I was awarded to participate in ACAL45. My heartfelt gratitude also goes to my colleagues Biniam Jembere
and Haile Gezae, both from Addis Ababa University, for their valuable comments and
suggestions on this paper.
36
At least, the three language model forwarded by Wodajo (2014), the replacement of Amharic by English as
the sole WL suggested by Gebreselassie (2015), etc., should all be considered, discussed and evaluated by
political groups to come up with better LPs with well-designed implementation strategies.
79
Mehari Zemelak Worku
Abbreviations
AEDP
AEUP
ANDO
ARDUF
BGDUF
BMPDO
DDPDP
DDQPDO
EDP
EJDFF
EPAP
EPLF
EPRDF
EPRP
ERAEIP
FDRE
FORUM
GPRDF
80
All Ethiopian Democratic
Party
All Ethiopian Unity Party
Argoba Nation’s
Democratic Front
Afar Revolutionary
Democratic Unity Front
Benishangul Gumuz
Democratic Front
Bahrworq Mesmes People’s
Democratic Organization
Dubie and Degeni People
Democratic Party
Denta, Dubamo
Qinchilchila People’s
Democratic Organization
Ethiopian Democratic Party
Ethiopian Justice and
Democratic Forces Front
Ethiopian Pan-Africanist
Party
Eritrean People Liberation
Front
Ethiopian People’s
Revolutionary Democratic
Front
Ethiopian People’s
Revolutionary Party
Ethiopian Vision Party
Federal Democratic
Republic of Ethiopia
Forum for Federalist
Democratic Unity of
Ethiopia
Gambella People’s
Revolutionary Democratic
Front
GSPP
HNL
KPC
MEISON
Geda System Proponents
Party
Harari National League
Kembata People’s Congress
OLF
All Ethiopian Socialist
Movement
Oromo Liberation Front
OLP
Oromo Liberation Party
ONC
Oromo National Congress
OPDU
Omo People Democratic
Unity
PDRE
People’s Democratic
Republic of Ethiopia
Provisionary Military
Administrative Council
Southern Nations
Nationalities and People’s
Region
Tigray People’s Liberation
Front
United Ethiopian
Democratic Forces
PMAC
SNNPR
TPLF
UEDF
UJDP
WPDF
WPE
WSDP
Unity for Justice and
Democracy
Wolaita People Democratic
Front
Workers’ Party of Ethiopia
Western Somalia
Democratic Party
5 How multilingual policies can fail
References
Awgchew, Girma. 2009. The origin of Amharic. Addis Ababa: Addis Ababa University
Press.
Balsvik, Randi Ronning. 2007. The quest for expression: State and University in Ethiopia
under Three Regimes, 1952- 2005. Addis Ababa: Addis Ababa University Press.
Bender, M. Lionel. 1976. Introduction. In M. Lionel Bender (ed.), The non-Semitic languages
of ethiopia, 1–24. East Lansing: Michigan State University in Cooperation with Southern Illinois University.
Berhanu, Kassahun. 2003. Political culture and party politics in Ethiopia. In Mohammed
Salih (ed.), African political parties: Evolution institutionalization and governance, 1–2.
London: Pluto Press.
Board, National Electoral. 2009. Political parties for election 2010. http://www.NEBE.net.
et, accessed 2011-09-11.
Bogale, Berhanu. 2009. Language determination in Ethiopia: What medium of instruction? In Svein Ege, Harald Aspen, Birhanu Tefarra & Shiferaw Bekele (eds.), Proceedings of the 16th International Conference of Ethiopian Studies, 1089–1101. Trondheim:
Norwegian University of Science & Technology.
Bowen, Donald & Nancy Horn. 1976. Language education. In Lionel Bender (ed.), The nonSemitic languages of Ethiopia, 608–634. East Lansing: African Studies Center Michigan
state University.
Bulcha, Mekuria. 1998. The politics of linguistic homogenization in Ethiopia: Trends in
the development of the oromo language. In Kwesi Kwaa Prah (ed.), Between distinction
and extinction, 1–2. Johannesburg: Witwatersrand University Press.
Chiswick, Barry. 2008. The economics of language: An introduction and overview. Chicago:
Institute for the Study of Labor.
Cobarrubias, Juan. 1983. Ethical issues in status planning. In Joshua A. Fishman & Juan
Cobarrubias (eds.), Progress in language planning: International perspectives. 41–87. Berlin:
Walter de Gruyter.
Cooper, Robert L. 1976. Government language policy. In M. L. Bender, J. D. Bowen, R. L.
Cooper & C. A. Ferguson (eds.), Language in Ethiopia, 187–190. Oxford: Oxford University Press.
FDRE. 1994. Education and training policy. Addis Ababa.
Fishman, Joshua A. 1972. Language in sociocultural change. Stanford: Stanford University
Press.
Gebreselassie, Fisseha. 2015. Choosing a working language in mutilingual nations: Rethinking Ethiopia’s working language policy. no address: no publisher. http : / / www .
aigaforum.com/articles/Paper- on- Ethiopia- Language- Policy.pdf, accessed 2015-0625.
Getachew, Anteneh & Derib Ado. 2006. History and current trends. Ethiopian Journal of
Education and Sciences 2. 37–60.
Grin, Francois. 1994. The economics of language: Match or mismatch? International Political Science Review 15(1). 25–42.
81
Mehari Zemelak Worku
Gudina, Merera. 2006. Contradictory interpretations of the ethiopian history. In David
Turton (ed.), Ethnic federalism: The ethiopian experience in comparative perspective, 119–
131. Addis Ababa: Addis Ababa University Press.
Heugh, Kathleen, Carol Benson, Berhanu Bogale & Mekonnen Alemu Gebre Yohannes.
2007. Final report study on medium of instruction in primary schools in Ethiopia. Addis
Ababa: MOE.
Leyew, Zelealem. 2012. A historical and typological overview. Ethiopian Journal of Languages and Literature 12(2). 1–59.
McNab, Christine. 1988. Language policy and language practice: Implementation dilemmas
in Ethiopian education. University of Stockholm dissertation.
McNab, Christine. 1990. Language policy and language practice: Implementing multilingual literacy education in Ethiopia. African Studies Review: Special Issue for the International Year of Literacy 33(3). 65–82.
Mekonen, Walelign. 1969. On the question of nationalities in Ethiopia. no publisher. http:
//www.walelignfordemocracia.com/onationalqu.pdf, accessed 2014-11-09.
Obeng, G. Samuel & M. Tristan Purvis. 1999. Sub-Saharan africa. In Joshua A. Fishman
(ed.), The handbook of language and ethnic identity, 334–390. Oxford: Oxford University
Press.
Pankhurst, Richard. 1976. Historical background of education in Ethiopia. In Lionel Bender (ed.), The non-Semitic languages of ethiopia, 305–323. East Lansing: Michigan State
University in Cooperation with Southern Illinois University.
Patten, Allan. 2001. Political theory and language policy. Political Theory 29. 691–715.
Philipson, David. 2012. Foundations of an African civilization: Aksum & the northern horn,
1000 BC-AD 1300. Addis Ababa: Addis Ababa University Press.
Revised Constitution of Ethiopia. 1955. Addis Ababa.
Schiffman, Harold. 1996. Linguistic culture and language policy. London: Routledge.
Smith, Lahra. 2008. The politics of contemporary language politics in Ethiopia. Journal
of Developing Societies 24. 207–243.
Stalin, Joseph. 1913. Marxism and the national question. http://www.marxists.org, accessed 2014-11-09.
Stoddart, John. 1986. The use and study of English in ethiopian schools. Addis Ababa.
Tzadua, Paulos. 1968. Foreward. In Peter L Strauss & Paulos Tzadua (eds.), The fetha nagast: The law of the kings, xv–xxix. Addis Adaba: Faculty of Law, Haile Sellassie I University.
UNESCO. 2015. Interactive atlas of the world’s languages in danger. no address. http://
www.unesco.org/languages-atlas/index.php, accessed 2015-06-25.
Vaughan, Sarah. 2006. Responses to ethnic federalism in Ethiopia’s Southern Region. In
David Turton (ed.), Ethnic federalism: The Ethiopian experience in comparative perspective, 181–207. Addis Ababa: Addis Ababa University Press.
Wodajo, Mesfin. 2014. Sociolinguistic challenges of the Post-1991 Ethiopian Language
Policy. Journal of Languages and Cultures 5(2). 17–23. http://www.academicjournals.
org, accessed 2015-06-25.
82
5 How multilingual policies can fail
Zemelak, Mehari. 2011. The language issue among Ethiopian political parties. Addis Ababa.
MA thesis unpublished.
Zewde, Bahru. 2002. A history of Modern Ethiopia, 1855-1991. Addis Ababa: Addis Ababa
University Press.
Zewde, Bahru. 2008. The changing fortunes of the Amharic language: Lingua franca or
instrument of domination? In Society, state and history: Selected essays, 77–95. Addis
Ababa: Addis Ababa University Press.
83
Part II
Language endangerment
Chapter 6
Linguistic imperialism and language
decolonisation in Africa through
documentation and preservation
Kofi Agyekum
The University of Kansas
This paper addresses the politics of language use in African nations and societies. It highlights the role of power and economics in the choice of language. It discusses linguistic
imperialism and language shift, and how they lead to language endangerment. The paper
also discusses linguistic decolonization whereby societies resist linguistic domination and
endangerment and embark on language maintenance. It touches on the methods employed
in language decolonisation, namely language revitalisation, resistance, maintenance, documentation and preservation. Attention will be on lexicology, terminology and the role
of radio and TV. We argue that as a society tries to redeem itself from linguistic imperialism through decolonisation, certain stronger politico-economic factors push it back into
linguistic imperialism. We will find out that some of the indigenous people themselves kick
against language decolonisation. The paper hinges on the theoretical base of language endangerment. Examples are taken from African and Ghanaian languages with emphasis on
Akan.
1 Introduction and definition of linguistic imperialism
According to Phillipson (2009: 780) “linguistic imperialism is the notion that certain languages dominate internationally on others. It is the way nation-states privileged one
language, and often sought actively to eradicate others, forcing their speakers to shift to
the dominant language.” The expressions, actively eradicating and forcing speakers are
too strong because despite the policies employed by the colonial masters in their former colonies, they cannot be said to be actively seeking to eradicate African languages.
According to Mous (2003: 158) “Contrary to the situation in Latin America, in Africa,
colonial languages are not a major factor in language loss.” In the context of this paper,
our working definition and conception of linguistic imperialism is “a linguistic situation
where the indigenous people are gradually conscientised to shun their indigenous languages and adopt foreign languages because of the benefits they expect to derive from
Kofi Agyekum. Linguistic imperialism and language decolonisation in Africa through documentation and preservation. In Jason Kandybowicz, Travis Major, Harold Torrence & Philip T.
Duncan (eds.), African linguistics on the prairie: Selected papers from the 45th Annual Conference
on African Linguistics, 87–104. Berlin: Language Science Press. DOI:10.5281/zenodo.1251718
Kofi Agyekum
them. They are made to believe that their languages cannot be used in any transaction
in education, economics, science and technology but instead a foreign language is the
best.
In linguistic imperialism, there is a greater relationship between political and economic dependence. Even though many African and Asian countries have won political
independence, there is still economic reliance and dependence on the developed and industrial world. This has culminated in linguistic and economic neo-colonialism. The European languages are still the dominant languages in science and technology, medicine,
engineering and ICT. The linguistic dominance of the English language in science and
technology has marginalized other languages in this field.
Phillipson (2009: 780) further argues that “linguistic imperialism entails unequal resource allocation and communicative rights between people, defined in terms of their
competence in specific languages, with unequal benefits as a result.” Linguistic imperialism has deepened in African countries to the extent that many families in the cities
communicate with their children in English, French, Portuguese and Arabic, and the children cannot speak their mother tongue. People bear western names and have shelved
their African names. Attitudes towards African languages in school and at home are very
negative, and people are ashamed of speaking their languages.
Linguistic imperialism shows that language is one of the manifestations of European
imperial expansion in Africa, Asia and Latin America. After the European Struggle and
Partition of Africa, new languages of communication emerged patterned on the colonial
masters’ languages. Linguistic imperialistic ideologies glorify the dominant languages
like Arabic, French, Hebrew, Latin, English, as the languages of religion and civilization.1
2 Language shift and language endangerment
This section discusses the theoretical notions of language shift and language endangerment. In language contact, language imperialism leads to language shift, whereby a community of speakers effectively abandons one language and shifts to another (Garret 2006:
63).2 Language shift is the gradual replacement of the communicative functions of one
language by another which the user considers to serve the maximum linguistic, political
and social benefits (see Agyekum 2009). The basic determining factor in language shift
relates to particular benefits to be derived from the use of the target language; especially
economic benefits (see Mufwene 2006, 2002: 175, Mous 2003: 160). Mufwene (2006: 116)
states that “English is spreading around the world because there are more and more
people who hope to find better jobs, to travel, to distant places, to be read by more and
more scholars, etc.” In language shift, the functions of the minority language diminishes
1
Phillipson (2009: 782) posits that linguistic imperialism is the maintenance of injustice and inequality by
means of language policies, and it is strongly connected to policies in commerce, science, international
affairs, education, culture and the media.
2
O’Shannessy (2011: 83) states that “in language shift, members of the community stop speaking the precontact language habitually and mostly speak the post-contact language, which comes to be the language
of the next generation.”
88
6 Linguistic imperialism and language decolonisation in Africa
in relation to the more “prestigious and politically powerful language” whose functions
become expanded to cater for previous functions of the minor language. Speakers find
the dominant languages more advantageous in all spheres of their lives. We can say that
language shift comes from two sources (1) linguistic imperialism involving forcing or
conscientisation, and (2) speakers making the change from the indigenous language to
another language whether a foreign or a major indigenous language for socio-economic
benefits.
Pandharipande (2002: 213–214) adopts the term functional load and asserts that the
higher the functional load, the more powerful the language is perceived to be. An increase in the functional load implies the promotion and sustenance of the dominant language (colonial or national language) while the trend of “linguistic diminishing returns”
in the functional load is serious. An uncontrolled language shift can lead to language
decay, endangerment, extinction and death or attrition. Derhemi (2002a: 151) avers that
“An endangered language is a language that may soon vanish, ceasing to be used as a
vehicle of communication, perhaps even disappearing completely from human history.”
In Ghana, the Togo-Mountain languages can be considered as examples of endangered
languages.
Language shift and death are engineered by globalization, language ideologies, language policies, language attitudes and everyday communicative practices in bilingual or
multilingual communities. Theoretically, communities that have positive attitudes about
their own language and community try to embark on language revitalisation and maintenance. In looking at the positive attitude towards language survival and maintenance
Adegbija (2001: 307) posits that “ As long as speakers of a language have a deep stake in
its survival and a high emotional involvement and commitment to its existence, all the
language shift agents and triggers in this world will not be able to kill their resolve.” By
contrast, negative attitude to one’s language results in language shift, culminating into
language death. Minority languages can die when their speakers are indifferent and lack
the effective interventional means to protect them (see Derhemi 2002a: 249).
The other major causes of language shift and endangerment are social status of the
language, educational and cultural factors, power relations of the languages that are in
contact, economic, migratory and settlement patterns, exogamous marriages, religion,
language policies and globalisation (Austin & Sallabank 2011: 5–6).
2.1 Language shift based on globalisation and international language
policies
Some language shifts are caused by globalisation while some are based on language policies coming from colonial masters or national governments.
2.1.1 Globalisation
Globalisation is one of the key factors of language endangerment (see Austin & Sallabank
2011: 21). In language shift within the context of globalisation, very few languages benefit
from international trade and international exchange. Brenzinger (2009: 446) asserts that
89
Kofi Agyekum
“The international exchange of knowledge and world trade are conducted in only a few
world languages, and some scholars, seem to expect that, given this tendency, a world
culture, based on one common language will finally emerge from these development.”
In the language of globalisation, the selected languages have wider communication and
they are also easily accessed (see Bamgbose 2011: 5).
Some imperial languages are used as supra-regional or global languages of communication in international organisations like UN, Council of Europe, NATO, EU, AU, and
ECOWAS. Imperial languages are used in negotiations between two or more states, languages of diplomacy, international agreements and pacts. English is increasingly becoming the world’s international lingua franca.3 Mous (2003: 160) records that “In Nigeria
there is the rebirth of Nigerian Pidgin English as a consequence of globalisation. Other
examples of language rebirth under globalisation in Africa are cases of urban slang losing
its stigma and serving the function of bridging ethnicity and taking over all communication situations in the big city.” Examples are Iscamto in Johannesburg, South Africa, and
the use of Ghanaian Pidgin English in secondary schools, universities and public work
places in Ghana.
Linguistic imperialism and education, and the expansion of the English language have
provided a market and economic force for the teaching of English as a second language
for foreigners. The English Language Teaching and Learning business is a major pillar
for the British economy. 4 English language has become a global commodity sold in the
international market. The socio-economic factors in language shift to a language like English for economic integration, cultural shifts, international networks and employment
makes language a more marketable commodity than a marker of identity (see Austin
& Sallabank 2011: 21). At the University of Ghana, students from Equatorial Guinea are
sponsored heavily by their government to learn English at the Language Centre. At the
department of Linguistics, Legon, a full-fee Sandwich Masters programme in TESL is
oversubscribed. The situation is different for an MA sandwich in Ghanaian Languages
at the University of Education Winneba. Very few students are ready to pay full-fees to
study Ghanaian languages. The Presbyterian University of Ghana has a Ghanaian languages department but has not been able to enroll a single student to pay full fees.
2.1.2 Language policies
In some cases legislation is passed declaring that a language is the only one that individuals are allowed to use in both public and non-public domains (see Lewis 2013: 677). Some
aspects of language shift deals with language of official communication. Most African
countries use the former colonial masters’ languages as the official languages in national
administration, governance, education, legislation, judiciary, media, etc. Politically, very
few of the masses in African countries speak the colonial languages; it is limited to the
3
The integration of the new European Union member states accelerates the spread of English as the European lingua franca.
4
Phillipson (2009: 781) records that the British economy benefits by £11billion directly and a further £11billion
indirectly from their intake of foreign students in Higher education. He continues that over half a million
foreign students attend language schools in Britain each year, spending time in learning English.
90
6 Linguistic imperialism and language decolonisation in Africa
few elites. For instance, in Ghana, the official parliamentary language is English and
nobody can become an MP without speaking English. Undoubtedly, no matter how brilliant one is, if s/he cannot communicate in English s/he cannot take part in any serious
and meaningful political deliberations, and cannot be a parliamentarian. What happens
in parliament is an evidence of disenfranchising of the masses by a few small classes
of educated elites (Bamgbose 2000). In Ghana, all formal and official interactions at the
governmental levels are done in English. The constitution and public documents are all
in English.
Within the same nation, some languages are instrumentalised by governments, and
they receive official and national support and recognition. In North Africa, Arabic varieties are expanding very fast. Kiswahili is established as a nationwide language for communication in Tanzania, and is threatening more than 130 other Tanzanian languages
(see Mous 2003: 159). The same phenomenon prevails in Botswana where Setswana language is dominating about 30 other languages in the country.
2.2 Education and language shift
Language shift occurs due to the type of language policy in education (see Agyekum
2009). Where a different language (colonial or African) is the language of education there
is a shift from the indigenous languages. In some schools in the urban centres in Ghana,
children from other indigenous languages have no option but to use, study and pass in
the local language. For example, all pupils who attend school in the Ashanti region study
Asante Twi for their Basic Education Certificate Examination (BECE). In the schools in
Accra, the capital city of Ghana, pupils are exposed to only three languages, namely Twi,
Ewe and Ga. Here too, the Twi language dominates and most children from the three
northern territories are forced to study Twi. This is subtractive bilingualism and can lead
to loss of self-confidence and lower achievements (see Austin & Sallabank 2011: 9).
In Ghana, one cannot enter any high school or tertiary institution without passing
English language. Many families thus forbid their children to learn or communicate in
the Ghanaian languages even at home so as to be fluent in English. Most private basic
schools prohibit their pupils from speaking any of the Ghanaian languages apart from
having them as courses on the timetable. Pupils who flout this rule are punished. Teachers hang placards around pupils’ necks reading, “I have spoken vernacular”, or “Vernacular is prohibited”. These trample on the pupils’ linguistic human rights and freedom of
speech.5
2.3 Media and language shift
Linguistic imperialism goes with media imperialism. Many news agencies in Africa carry
their information mostly in English, French or Portuguese. Politically, the world’s infor5 Adegbija (2001: 285) records that “Ex-colonial languages have been functionally placed on a pedestal, being
attitudinally extolled and being seen as inevitable both at the individual and societal levels.” He further
stated that “In Nigeria, Ghana, Uganda, Kenya and many other African countries colonized by the British,
for instance, to be educated is virtually synonymous with knowing and being able to use English.”
91
Kofi Agyekum
mation channels are dominated by these few languages, and to follow the global trends of
information, one needs to know one or more of these languages.6 Most programmes, classified advertisements on jobs, funeral announcement, conferences and other important
announcements in the print media, radio and TV are mainly in the colonial languages.
There is a social stratification based on the degree of access to communication where
speakers of the indigenous languages are not able to use their own languages in broadcasting. They are compelled to use the dominant languages, and this shift can lead to language endangerment and marginalisation of the masses. In Ghana, after independence,
there were a few newspapers in the local languages, including, Nkwantabisa in Asante
Twi and Ewe. Unfortunately, there is not a single newspaper printed in any Ghanaian
language now, The Daily Graphic, the most widely circulated Ghanaian newspaper tried
a pull up in its Tuesday edition and captioned it Ghana Nsɛmpa ‘Ghana’s Good News’ but
it did not go well so they have stopped. Until the proliferation of many FM radio stations
that broadcast in the local languages, many people who listened to Ghana Broadcasting
Corporation (GBC), BBC, and VOA were elites. The current use of indigenous languages
in radio and TV is a situation of language decolonisation where the media shields the
local languages from linguistic imperialism. In Accra most FM radio stations broadcast
in Akan, and the local language Ga, was succumbing to the Akan language until the government established Ogbonu FM that broadcasts only in Ga to shield the Ga language.
2.4 Religion and language shift
Every religion has a political undertone especially when it is a foreign religion. It can
change the ideology and attitude of the adherents. Adherents to new religious faiths must
learn the language associated with it, and shift from their indigenous languages during
worship (Brenzinger 2009: 447–448). For example, Arabic, Hausa and Dyula languages
have spread in West Africa along with Islam. All good Moslems, especially the Sheiks,
are expected to be proficient and knowledgeable in the Arabic language and culture.7
2.5 Economic occupational and migratory language shift
Economic and occupational survival are crucial factors in language learning and shift
either from a minority to a majority language or vice versa (see Salzmann 2004: 194,
Holmes 2001: 58).8 Africans in Europe, Asia and American learn the host language so as
to be employed. Theoretically, notwithstanding how strongly people are attached to their
languages and cultures, the socio-economic pressures in modern global world make it
6
Unequal benefits in language also lead to cultural imperialism where one society spreads its culture unto
others such that they will throw away their cultures. Most African countries have fallen into this trap.
7
According to the literature, Arabic was strongly attached to the widespread of Islam because people firmly
believed that it was very difficult if not impossible to translate the Quran into another language (see SavilleTroike 1989: 208). This myth of the intranslatability of the Quran has now been broken since there are
English and Akan versions of the Quran.
8
Philips (2006: 485–486) asserts that ‘The political economic position of a group determines its attitude
towards the codes in a group’s multilingual repertoire, the group’s code choices, and the ultimate survival
of the codes being spoken. The inequality of languages originates in economic equality.’
92
6 Linguistic imperialism and language decolonisation in Africa
difficult for them to glue themselves to ancestral languages and culture, they have to shift
to another language (see Mufwene 2002: 190). Lack of language and cultural practices
gradually lead to language attrition and death.
Economic factors can raise a language to the status of a lingua franca and many nonnative speakers will have the urge and zeal to learn and speak it. This supports an assertion by Mufwene (2006: 130) that “A particular language can be preferred because
it functions both as vernacular and as a lingua franca, or because it is associated with
money- making or better paying jobs, or because it affords its speakers a better social
status, or because it is more widely spoken demographically and/or geographically.”
Adegbija (2001: 286) asserts that “The more powerful and functionally the dominant
language, the more pressing its attraction and pull and the greater the tendency to shift
towards it given the fact that the pressure for social vertical mobility is virtually irresistible.” In view of this notion, many dominant languages in Africa are replacing the
smaller ones. Examples are the spread of the Somali language in Somali, Bamana or Jula
in Mali, and Burkina Faso, Hausa in Nigeria and Niger, Yoruba and Igbo in Nigeria, Lingala in Congo, and Wolof in Senegal, Kikuyu in Kenya, Akan and Ewe in Ghana. These
replacement are normally based on urbanisation, economic benefits and prestige (see
Mous 2003: 160, Adegbija 2001: 286)
In Ghana, the Akan language is dominating the other indigenous languages very fast.9
Non-Akan speakers naturally find it more advantageous to learn Akan so as to operate
easily and extensively in Ghana. It is used mostly in the non-Akan cities and major towns
like Accra, Bolgatanga, Yendi, Tamale, Navrongo, Wa, as the preferred lingua franca in
religion, trade and commerce, music and the arts, artisanship and many aspects of the
society. The same situation prevails in the Volta region where most of the Togo Mountain
minority languages are being dominated by the Ewe language. In the Akan and Ewe
examples, we can argue that the language shift from the minority languages is not caused
by Indo-European languages but by African languages that the groups choose based on
socio-political benefits (see Mous 2003, Brenzinger 1998, Lüpke & Storch 2013).
Social mobility, migration and settlement patterns result in language contact and language shift (see Agyekum 2009). Some people move to urban centres and abandon their
languages for the ‘so-called’ prestigious ones. However, large groups of immigrants often manage to preserve their languages.
3 Language decolonisation
The term language decolonisation is derived from colonisation and decolonisation. Jaffe
(2009: 534) states that
Linguistic decolonisation (LD) describes both the actions taken in postcolonial contexts to undo the social, political, and cultural effects of the dominance of colonial
9
In Ghana, Akan is spoken in six out of the ten regions and by 47.5% of the entire population as native
speakers and also by 44% of the remaining Ghanaians as non-native speakers.
93
Kofi Agyekum
languages and a philosophical challenge to the Western language ideologies that
underpinned the colonial project and that have persisted in the postcolonial period.
3.1 Strategies and tools for language decolonisation
Language decolonisers employ documentary approaches, sociological, ideological and
pragmatic concepts to fight against language shift, endangerment, extinction and death.
In language decolonisation, states and societies take nationalist interventions, efforts
and action plans to legitimate their languages and identities that have been suppressed
by linguistic imperialism.
They use multifaceted avenues and tools for language revitalization, and maintenance.
These include educational curriculum, literacy and awareness of language death, media,
especially radio, linguistic pluralism, language documentation and preservation. Language decolonisation, documentary projects and interventions redress the linguistic inequality, manipulation, repression and cultural oppression created by the replacement
of the functions of their language by the majority and colonial languages.
3.2 Language decolonisation through radio, TV and pop-culture
The media can be a strong tool for language decolonisation; it serves as a platform for
the use, documentation and development of the endangered languages.10 To Moriarty
(2011: 447) ”Endangered media and pop-culture can help raise the status of the relevant
language, aid corpus planning through the dissemination of new terminology, and encourage language acquisition by increasing language exposure in both the public and
private domains.” In Ghana, the mass media, especially radio record most of their programmes in the Ghanaian languages. Mass media serves as a mechanism for the storage
of expressions, reservoirs and reference points for the circulation of words, phrases and
discourse, proverbs and other aspects of Ghanaian language and popular culture (see Debra 2001: 96). The media as a tool for language decolonization and language maintenance
serves as an unplanned and surest area where language changes and practical language
policies can be employed spontaneously by the speech community (see Agyekum 2010,
Derhemi 2002b: 158, Moriarty 2011: 447).
3.3 Recycling, recontextualisation and reinterpretation of media
discourse
Oral media discourse in Africa is an essential aspect of the people’s life and a new social
interconnection across socio-structural, ethnic, religious, professional and party groupings. African societies have public and private FM stations that have adopted African
languages policy. Some do 50%, 70% and even 100% African languages programmes. In
Ghana, examples are Ogbonu FM in Tema (Ga), Ada FM (Ada), Peace FM. Nhyira, Adom,
10 Debra
(2001: 95) records that “there are several cases in which phrases and discourse styles are extracted
from radio broadcasting and then recycled and reanimated in everyday usage outside of the contexts of
radio listening”.
94
6 Linguistic imperialism and language decolonisation in Africa
Oman, and FM, (Akan). These policies are means of revitalising, maintaining, preserving
and documenting their languages as part of the process of language decolonisation.
3.4 Linguistic impact
Linguistically, African language programmes on radio and TV have immensely contributed in African societies. Listeners learn a lot of things on language such as new
vocabularies for modern concepts in politics, medicine, health, education, administration, economics and science from African programmes. They get new terms, idiomatic
expressions, etymologies of words, proverbs, archaisms, appellations, etc. of indigenous
languages (see Agyekum 2010 on radio).
One of the major functions of the African language programmes on radio and TV
talk-shows is language modernization, development and elaboration of terms to cater
for most aspects of human life. Radio is one of the most powerful tools in the dissemination, interpretation and recontextualisation of discourse (see Agyekum 2000, 2010).
Coined terms and phrases commonly used on African language radio programmes are
picked up by the people and accepted for use outside radio. Below are some Akan (Ghana)
terms popularly used, they include:
(1)
dumsɔ is from dum, ‘to switch off’ and sɔ, ‘switch on’, implying ‘power outage’.11
(2) ɔmampanin, is made up of ɔman, ‘state’, panin, ‘elder’ and hence ‘president’.
(3)
ɔsoafoɔ, ‘minister’ is made up of soa, ‘to carry’ and agentive prefix ɔ- and suffix
-foɔ. The minister is the carrier of the “load” and responsibility of the ministry.
(4) mmarahyɛbadwa, ‘parliament’, the components are mmara, ‘laws’, hyɛ, ‘to fix’
and badwa, a gathering of people who are tasked to make laws.
(5) ankorɛankorɛ kyɛpɛn, for ‘human rights’ the adjective korɛ means singular and
the derived compound ankorɛankorɛ means individuals, kyɛpɛn is made up of
kyɛ, ‘to share’ and pɛn means level or equal. The composite term thus means
equal distribution of entities to people (see Agyekum 2006 for other terms).
African oral literature genres appear on radio because they are less available in urbanized areas where the heterogeneous nature of the society is making people lose their cultural heritage. The radio and TV are becoming repositories (and innovators) for African
languages especially in areas of orality. Radio culture is gradually replacing the traditional “village” oral traditions that were provided by the elders. Until the advent of the
missionaries, the African languages were purely oral. Storytelling and proverb sections
that were earlier on provided by the elders in the villages are now done effectively on
African radio and TV programmes for a wider audience (see Agyekum 2000). The air
waves allow us to reach “the villages”, the radio is wider than the reach of the village
oral literature sessions.
11
Dumsɔ, ‘power outage’ has gained international recognition and has been used in Wikipedia and in Google.
95
Kofi Agyekum
3.5 Linguistic pluralism and official multilingualism as language
decolonisation
One of the surest theoretical approaches to language maintenance and decolonization in
African multilingual societies is the policy of linguistic pluralism and official multilingualism whereby various languages are decolonised and liberated from language imperialism
and endangerment. Linguistic pluralism or official multilingualism is a system that promotes the coexistence of different languages and allows them to operate on equitable
basis. The same opportunity for development and usage are given to more languages
irrespective of their number of speakers. Linguistic pluralism is a democratic way of
dealing with linguistic varieties in a multilingual nation.
Eritrea is one of the recognised linguistic pluralistic nations in Africa that gives all
the nine indigenous languages the same opportunity in the mass media and in schools.12
Societies that practise official multilingualism include Canada (French, English) Switzerland (French, Swiss, and German) and South Africa (11 languages).13
The most obvious problem with linguistic pluralism is economic, we use more money
for printing books, training personnel, employing examiners, etc. A cost benefit analysis
will prove that money spent in operating linguistic pluralism would be far less than
money spent in peacekeeping if conflicts and wars emerge out of linguistic imposition
and imperialism.
3.6 Language resistance and maintenance
Let us continue our discussion with language resistance and maintenance as tools of language decolonisation. It is possible for a community to undertake language revitalization
and maintenance during the time that the shift is taking place. This happens in a situation where the speakers of the language undergoing a shift are loyal to their language
and are bent on maintaining it. Nahir (2003) states that
Language maintenance is the preservation of the use of a group’s native language,
as a first or even as a second language, where political, social, economic, educational, or other pressures threaten or cause (or are perceived to threaten or cause)
a decline in the status of the language as a means of communication, a cultural
medium, or a symbol of group or national identity. (Nahir 2003: 439).
Lewis (2013: 673) also posits “Language maintenance is the effort to arrest and reverse
the process of language shift: an effort is made to ensure that a vulnerable language does
not decline and eventually disappear, but rather it continues to be spoken by a sustainable community of people.” Language maintenance is the situation where a language
group that is bound to shift to other languages and become endangered under linguistic
12 The
nine languages used in Eritrea are Afar, Arabic, Beja, Blin, Kunama, Nara, Saho, Tigre and Tigrinya.
Ghana embarks on partial linguistic pluralism, and offers eleven languages in schools at the Junior High
School (JHS), Senior High School (SHS) and universities.
13
The South Africa 1996 Constitution has a new language policy that accords official status to eleven languages also referred to as ‘de facto’ national languages spoken by major populations within the country.
The aim is to avoid linguistic conflicts.
96
6 Linguistic imperialism and language decolonisation in Africa
imperialistic circumstances, rather holds to their language and expands its functional
load (see Coulmas 2005: 157).
Most Africans adhere to various practical ways to maintain their languages and culture. Culture and language are inextricably interwoven. The loss of a language through
language shift and endangerment logically results in the loss or reduction of cultural
practices (Derhemi 2002b: 153). These include oral literature genres like proverbs, folktales, myths, legends, historical narratives, folk songs and poetry (see Crystal 2000).
There are good reasons to protect languages and cultures from dying.
Speakers of a particular language group themselves can maintain their language instead of shifting to the dominant language. An extended form of this type of language
maintenance is found in African communities in Europe and North America. African
ethnic groups congregate from different parts of the cities in which they live to meet
regularly and chat in their languages, eat African food, and wear African clothes. They
have African musical groups that meet occasionally to perform. They have chiefs, queens
and sub-chiefs, and they continue to practice their cultures and languages. They have established African churches where the worship is conducted in African languages and
they sometimes invite African pastors from Africa to preach. These types of gatherings
help to promote African languages and cultures in the developed world.
4 Language documentation
This section discusses basic concepts, techniques, and challenges of language documentation.
4.1 Basic concepts about language documentation and preservation
Let us finally look at decolonising language through documentation and preservation of
aspects of the language including the phonology, syntax, semantics, ethnography, culture, and oral literature.
To salvage situations of language death and endangerment, linguists counteract language death through language preservation, revitalization and documentation. The linguistic documentation of dying languages is sometimes labelled as “salvage linguistics”
(Craig 1997: 257). The documentation of language and oral art is the major technique that
acts as the catalyst for its preservation and language decolonization, because it is a way
for putting out materials in the form of books, CD Rom, and all kinds of electronic forms.
These sources can be consulted for specific aspects of language teaching, learning and
research.
Language documentation is a proactive process and involves actions that are meant
to record, maintain and preserve a language for the future; it is also geared towards
protecting it from language shift and death (see Agyekum 2012). From this standpoint,
language documentation precedes preservation.14 To Himmelmann (2006: 1), “Language
14
Trilsbeek & Wittenburg (2006: 3140) identify three agents in the archiving of language and these are (a) the
depositors, (b) the users, and (c) the archivists. The depositors record, and create different sorts of material
and hand these over to the archivists (documentation). The users are those who use the materials, and the
archivists solve the long-term preservation problems (preservation).
97
Kofi Agyekum
documentation is a lasting, multipurpose record of language.” The researcher should rely
on the linguistic knowledge of the native speakers on flora and fauna, kinship systems,
artefacts, food items, religion, narratives and oral literature genres, etc. If native speakers are actively involved in the creation of the document, the researcher gets firsthand
information about the language, and this increases the authenticity of the information
for documentation and preservation.
Before independence, language documentation in Ghana was carried out by missionaries like the Presbyterian and the Methodist who also translated the Bible into Twi, Fante,
Ewe and Ga. Since independence, documentation in the country has been undertaken by
the Bureau of Ghanaian Languages, Ghana Publishing Corporation and Ghana Institute
of Literacy and Bible Translation. The Linguistic and Ghanaian language departments
of the universities research into Ghanaian languages by producing long essays, theses,
dissertations, primers, dictionaries and journal articles (see Agyekum 2012).
4.2 Techniques in language documentation
The major strategies of language documentation are the recording, processing, preservation and dissemination of the primary data (see Woodbury 2011: 159). It involves audio
and/or video recording of all communicative events such as folktales, myths, proverbs,
folksongs, dirges, appellations, ordinary conversations and all other observable linguistic behaviour. In documentation, we collect the language materials and put them into
archival database, and disseminate the information for public use (see Salffner 2006: 106).
The scope of language documentation covers a large set of primary data that provides
strong evidence of language use in their natural sociocultural settings such as funerals,
traditional rituals, greetings, requests, apology, thanking, political discourse and chieftaincy (see Agyekum 2012). The language documenter elicits information from people
and takes notes from earlier documents like literature books, history and religion or
recorded court proceedings and notes from the Bible where they exist.
Language documentation should include many varied records and everything that is
related to language. To Himmelmann (2006: 3), “The goal of language documentation is
to create a record of language in the sense of a comprehensive corpus of the primary data
which leaves nothing to be desired by later generations wanting to explore whatever
aspect of the language they are interested in.” No matter how thoroughly we record and
document it, there would be areas that are undocumented.
In specific subject language documentation, the researcher could set some limits based
on his focus and needs. There is the danger of producing what Himmelmann (2006: 4)
calls “data graveyards”. This refers to large heaps of data with little or no use to anyone.
However, if we collect very limited data, the documented information may not reflect
the real nature and state of the language under study (see Agyekum 2012).
Data is of no importance if it is not well formatted and catalogued for easy accessibility
(see Agyekum 2012: 26). Any data should be accessible to the compilers of the document,
98
6 Linguistic imperialism and language decolonisation in Africa
and also be user friendly and accessible to a broad range of users including children,
researchers, and foreigners who want to learn the language in future.15
4.3 Factors militating against language decolonisation and
documentation:
Most linguistic decolonisers do not have the requisite resources and materials needed
to develop and raise their languages to the levels that can replace the status and functions of the colonial languages in all domains. The colonial masters are politically and
economically powerful, and their languages resist the challenges of decolonization, and
they embark on official monolingualism (see Lewis 2013). Fishman (2001: 13) states that
“the resources available to threatened languages are often quite meager and constantly
fewer than those available to their Big Brother rivals and competitors…. Threatened languages often have no outside support of any optimal significance to fall back upon.” For
instance, at the University of Ghana, most of the funds for research are granted to the
sciences and social sciences and not to Linguistics, Languages and Performing arts.
Another reason for the resistance to Language Decolonisation is that language planning policies favour foreign languages and majority national languages; this phenomenon makes language decolonisation difficult. This unfortunate situation prevails in Ghana where the official language policy states that the medium of instruction for the first
three years should be the indigenous Ghanaian languages, but the “international and
preparatory” schools in the urban areas rather use the English language as the medium
from day one especially when the pupils come from diverse linguistic backgrounds.
The development of the national and the minority languages for education and other
modern domains creates a division between the pure and mixed codes of the various
languages (see Jaffe 2009: 534). If the minority and threatened languages are only taught
as school subjects and not used in homes there is a disconnection, and that retards the
progress of the language. This is what normally happens in the urban centres, where the
minority language becomes a second language (see also Fishman 2001: 14–15). The use
of the colonial languages brings about elitism and social stratification that help the elite
to lord over the non-elite. They therefore want to do everything possible to thwart the
efforts of the linguistic decolonisers.16
Some scholars argue that language preservation and documentation are very costly
and not cost effective. Mufwene (2002: 162, 191) asserts that language development should
take a natural path. The changes occurring in linguistic behaviour of speakers are simply
an adaptation to changes in the socio-economic conditions of the speech communities
15
The documents should provide information about; participants, purpose, setting (when and where the data
was recorded), contents, structure and quality of the data, general information about the speech community
and the language, the data collection methods, links and references to other resources (books and articles),
(Himmelmann 2006: 11–12). The materials should be stored in hardcopy, softcopy, backup, pen drives, CDs,
DVD, CD Roms, e-mail accounts, etc.
16 Jaffe (2009: 535) concludes that “Postcolonial linguistic agents are often faced with a double-bind: if they
use the colonial language, they are seen as traitors to their cultural/ethnic group; if they use the dominated
language, their voice has a more limited power and reach.”
99
Kofi Agyekum
driven by interests related to costs and profits that come along with language use, and
therefore must be respected. Theoretically, language shift, endangerment and death are
all part of language evolution and must be allowed to take their natural path based on
human behaviour.
Generally, if speakers of an endangered language have more pressing socio-economic
concerns and prospects to think about, they are less worried about the fates of their
languages and how to salvage them from linguistic imperialism. Despite this conundrum
facing Language decolonisation, it is prudent to forge ahead since linguistic imperialism
tramples on peoples’ “linguistic human rights”, and language endangerment is a machine
that drives languages into the morgue.
Problems that militate against research, documentation, and publication of indigenous
languages include lack of cooperation from the informants, especially on issues relating
to the secret elements of oral art, filming materials or the performance itself. Translating
oral literature materials is not an easy task, many words and concepts are archaic and
obsolete, some flora, fauna and particular names are untranslatable.
Many African universities have a negative attitude towards teaching and research
in African languages, oral literature and cultural studies, and will not provide funds to
promote the development of African languages. There are difficulties in finding outlets
and funds for publication of research materials in oral literature, lexicology and other
aspects of African languages.
4.4 Recommendations for the development of African languages
All languages should be given the opportunity to survive so as to maintain their cultural
heritage (see also Adegbija 2001: 305–306). Based on the points raised in this paper, we
make the following recommendations and conclusions for the promotion, development
and decolonisation of African languages.
• The provision of vocabulary and appropriate terminology for the teaching and
learning of African languages should be made an important aspect of language development. To cope with language modernisation and elaboration, lexicology and
terminology should be integral parts of the language curriculum at the teachers’
training colleges, language centres and the linguistics departments of our universities. This will help us to produce word lists, terms and special dictionaries, glossary books on all aspects of African linguistics, English- African reference books
for African language students, broadcasters, translators, etc.
• We must develop and elevate the status of the African languages and literatures to
be used in education to teach our languages, mathematics, science, and technology
in all our schools, universities and technical institutions.
• At the training colleges, the teaching and research into African languages should
be intensified to produce well-trained knowledgeable and adequately qualified
teachers to teach our African languages (see Prah 1993: 9, Adegbija 2001: 305).
100
6 Linguistic imperialism and language decolonisation in Africa
• More books and articles should be written on African languages. The more we document and preserve the language, the better we are able to revitalise, maintain and
decolonize our African languages from linguistic imperialism. We could translate
some of the African, Caribbean, and Western literature into the African languages
to expand our textbooks and supplementary readers (see Agyekum 2003).
• A branch of language planning and modernisation should be established within
the Academy of Arts and Sciences in African countries. Our departments of languages at the universities should teach courses on language planning and policies
especially, MTE.
• African languages, linguistics and literature institutes should establish language
planning branches. We should also have national interdisciplinary lexicographical
and terminological centres to deal with language documentation in the various
African languages. Africans should have explicit, practicable, implementable and
firm language policies and planning on MTE rather than the theory and “skiphopping policies”.
• We should establish language media; radio, TV and newspapers and popular culture domains that the youth will be involved in and enjoy. The media can shape
and change their negative attitudes and ideologies about African indigenous languages.
• African agencies like the AU, ECOWAS and zonal groups should equip Bureaux
of African Languages in our countries with personnel, technical know-how, and
funds to develop our own languages (see Agyekum 2012, 2003). The 21st century
should be dubbed “Afrolingual Independence and Development”. We should be
proactive to develop our mother tongues and decolonise them from linguistic imperialism, language endangerment and death. The overall result of positive linguistic intervention will help our social, economic, religious, political, scientific, and
technological development.
5 Conclusions
In this paper, we have seen that language shift is part of language and politics and it
is found in language policy, politics, language of education, religion, media, economics,
immigration and settlement patterns. Naturally, a language with limited instrumental
utility to catch up with modernisation and globalisation is replaced with the one that
serves the current needs of the people. The international exchanges of knowledge and
world trade are conducted in only a few world languages. We opined that language imperialism occurs in two folds (a) conscientisation of a people to gradually shift from their
language, and (2) shift of a language for socio-economic and political benefits.
We paid attention to Linguistic decolonisation as a process where a society reinstates
its language status and reverses language shift. Decolonising agencies in African redeem
101
Kofi Agyekum
and maintain their languages across geographical frontiers. Unfortunately, while language decolonisers work very hard, globalisation, language attitude by some elites, modernisation and technological influence, borrowing and diffusion hamper their progress.
Problems relating to the collection, documentation and publication include the uncooperative and negative attitude of informants and speakers, lack of funds, problems of
translation, and difficulty in finding publishers (see Bamgbose 2011: 5).
Theoretically, every language can become a vehicle of modern civilization to meet the
demands of her people. The non-use of a language in both informal and formal set ups
contribute to language shift, endangerment, decay and loss, and restricts its functional
usage, but the documentation of a language is a big step in the process of language
survival and restoration (see Derhemi 2002a: 256).
The survival of every language theoretically depends on whether it is appropriate to
be used for all contexts of communication identified by the society such as marketing,
occupation, religion, music, etc. A language begins to lose its prestige and eventually
decays if its functional load is restricted to only informal usage. Our languages should
be integrated into ICT. It is only when our languages are able to cope with most of our
societal needs that Africa’s independence can be said to be complete.
Finally, the paper has argued that if we develop African languages, it will have a corresponding impact on African culture, art, politics, economics, agriculture, industrialisation, technology, medicine, primary health, commerce and trade, environmental studies,
media and education. If the modern ideas and information are disseminated in the indigenous African languages, their impact will be greater and help the majority of the masses
to understand them better. Political independence without linguistic independence is partial independence. In such a situation, our pace in modernisation will be slow, and we
will still be under linguistic imperialism and language endangerment (Agyekum 2003).
References
Adegbija, E. 2001. Saving threatened languages in Africa: A case study of oko. In Joshua
A. Fishman (ed.), Can threatened languages be saved? Reversing language shift, revisited:
A 21st century perspective, 284–308. Clevedon: Multilingual Matters.
Agyekum, Kofi. 2000. Aspects of akan oral literature in the media. Research Review 16(2).
1–18.
Agyekum, Kofi. 2003. Akan linguistics: Metalanguage and terminology. Legon: University
of Ghana dissertation.
Agyekum, Kofi. 2006. Lexical expansion and elaboration in akan: Afisem and the media.
Issues in Political Discourse Analysis 1. 71–85.
Agyekum, Kofi. 2009. Language shift: A case study of Ghana. Sociolinguistics Studies 3(3).
381–403.
Agyekum, Kofi. 2010. Ghanaian radio and the Akan language: Unplanned language planning and development. Issues in Political Discourse Analysis 3(2). 141–161.
Agyekum, Kofi. 2012. Documentation and preservation of the akan language. Basic Research Journal of Education Research and Review 1(2). 23–37.
102
6 Linguistic imperialism and language decolonisation in Africa
Austin, Peter K. & Julia Sallabank. 2011. Introduction. In Peter K. Austin & Julia Sallabank
(eds.), The Cambridge handbook of endangered languages, 1–24. Cambridge: Cambridge
University Press.
Bamgbose, Ayo. 2000. Language and exclusion: The consequences of language policies in
Africa. Munster: LIT Verlag.
Bamgbose, Ayo. 2011. African languages today: The challenge of and prospects for empowerment under globalization. In Eyamba G. Bokamba, Ryan K. Shosted & Bezza
Tesfaw Ayalew (eds.), Selected proceedings of the 40th Annual Conference on African
Linguistics, 1–14.
Brenzinger, Matthias (ed.). 1998. Endangered languages in Africa. Köln: Rüdiger Köppe.
Brenzinger, Matthias. 2009. Language maintenance and shift. In Jacob L. Mey (ed.), Concise encyclopedia of pragmatics, 2nd edn., 444–452. Amsterdam: Elsevier Ltd.
Coulmas, Florian. 2005. Sociolinguistics: The study of speakers’ choices. Cambridge: Cambridge University Press.
Craig, Colette Grinevald. 1997. Language contact and language degeneration. In Florian
Coulmas (ed.), The handbook of sociolinguistics, 257–270. Oxford: Blackwell.
Crystal, David. 2000. Language death. Cambridge: Cambridge University Press.
Debra, Spitulnik. 2001. The social circulation of media discourse and the mediation of
communities. In Alessandro Duranti (ed.), A companion to linguistic anthropology, 95–
118. Malden: Blackwell Publishers.
Derhemi, Eda. 2002a. The endangered arbesh language and the importance of standardized writing for its survival: The case of piani degli albanesi, sicily. International Journal of Multicultural Societies 4(2). 248–269.
Derhemi, Eda. 2002b. Thematic introduction: Protecting endangered minority languages:
Sociolinguistic perspectives. International Journal of Multicultural Societies 4(2). 150–
161.
Fishman, Joshua A. 2001. Why is it hard to save a threatened language? In Joshua A.
Fishman (ed.), Can threatened languages be saved? Reversing language shift, revisited:
A 21st century perspective, 1–22. Clevedon: Multilingual Matters.
Garret, Paul B. 2006. Language contact and contact languages. In Alessandro Duranti
(ed.), A companion to linguistic anthropology, 46–72. Malden: Blackwell Publishers.
Himmelmann, Nikolaus P. 2006. Language documentation: What is it and what is it good
for? In Jost Gippert, Nikolaus P. Himmelmann & Ulrike Mose (eds.), Essentials of language documentation, 1–30. Berlin: Mouton de Gruyter.
Holmes, Janet. 2001. An introduction to sociolinguistics. 2nd edn. Harlow, England: Longman Publishers.
Jaffe, Alexandra. 2009. Linguistic decolonialization. In Jacob L. Mey (ed.), Concise encyclopedia of pragmatics, 2nd edn., 534–535. Amsterdam: Elsevier Ltd.
Lewis, Huw. 2013. Language maintenance: A liberal-egalitarian approach. Journal of Multilingual and Multicultural Development 34(7). 672–689.
Lüpke, Friederike & Anne Storch. 2013. Repertoires and choices in African languages.
Berlin: de Gruyter.
103
Kofi Agyekum
Moriarty, Máiréad. 2011. New roles for endangered languages. In Peter K. Austin & Julia Sallabank (eds.), The Cambridge handbook of endangered languages, 446–458. Cambridge: Cambridge University Press.
Mous, Maarten. 2003. Loss of linguistic diversity in Africa. In Mark Janse & Sijmen Tol
(eds.), Language death and language maintenance: Theoretical, practical and descriptive
approaches, 157–170. Amsterdam/Philadelphia: John Benjamins.
Mufwene, Salikoko S. 2002. Colonisation, globalisation, and the future of languages in
the twenty-first century. International Journal of Multicultural Societies 4(2). 162–193.
Mufwene, Salikoko S. 2006. Language endangerment: An embarrassment for linguistics.
Proceedings from the Parasessions of the Forty-second Meeting of the Chicago Linguistic
Society 42(2). 111–140.
Nahir, Moshe. 2003. Language planning goals: A classification. In Christina Bratt
Paulston & G. Richard Tucker (eds.), Sociolinguistics: The essential readings, 423–448.
Malden: Blackwell Publishers.
O’Shannessy, Carmel. 2011. Language contact and change in endangered languages. In
Peter K. Austin & Julia Sallabank (eds.), The Cambridge handbook of endangered languages, 78–99. Cambridge: Cambridge University Press.
Pandharipande, Rajeshwari V. 2002. Minority matters: Issues in minority languages in
India. International Journal of Multicultural Societies 4(2). 213–234.
Philips, Susan U. 2006. Language and social inequality. In Alessandro Duranti (ed.), A
companion to linguistic anthropology, 474–495. Malden: Blackwell Publishers.
Phillipson, Robert. 2009. Linguistic imperialism. In Jacob L. Mey (ed.), Concise encyclopedia of pragmatics, 2nd edn., 780–782. Amsterdam: Elsevier Ltd.
Prah, Kwesi Kwaa. 1993. Mother tongue for scientific and technological development in
Africa. Cape Town: University of the Western Cape.
Salffner, Sophie. 2006. Learning endangered languages. A linguistic frame of reference
for modelling readers. In Francis Ogbokhare (ed.), Globalization and the future of
African languages, 104–116. Ibadan, Nigeria: Ibadan Cultural Studies Group.
Salzmann, Zdeněk. 2004. Language, culture and society: An introduction to linguistic anthropology, 3rd edn. Oxford: Westview Press.
Saville-Troike, Muriel. 1989. The ethnography of communication: An introduction. 2nd edn.
Oxford: Blackwell.
Trilsbeek, Paul & Peter Wittenburg. 2006. Archiving challenges. In Jost Gippert, Nikolaus
P. Himmelmann & Ulrike Mosel (eds.), Essentials of language documentation, 311–335.
Berlin: Mouton de Gruyter.
Woodbury, Anthony. 2011. Language documentation. In Peter K. Austin & Julia Sallabank (eds.), The Cambridge handbook of endangered languages, 159–186. Cambridge:
Cambridge University Press.
104
Chapter 7
Dictionary Day: A community-driven
approach to dictionary compilation
Bryan D. Gelles
University of Florida
A common component of language documentation is the compilation of a small dictionary.
The method of compilation has changed very little in the last century: most documentarians
elicit individual lexical items from a speaker and check the item through both translation and
backtranslation with other speakers. Two major problems with this method are the absence
of larger community engagement and idiosyncratic problems that come from lexical item
elicitation.
Animere is an endangered language spoken by around thirty speakers all aged over forty
years. The speech community is located in Kecheibi, northern Volta Region, Ghana. Over a
five month period I began the initial documentation of Animere with funds provided by a
Small Grant from the Endangered Languages Documentation Programme, integrating Dictionary Day, one day a week when members of the community would gather to discuss
lexical items. This method proved highly successful: I saved time and funds by making use
of the speech community’s intuition while obtaining valuable folk linguistic information
when there was disagreement. Furthermore, the speech community was not only engaged
but agentive, allowing for genuine consultation between the linguist and the speech community. The major drawback, however, is lack of synergy among documentarians and other
linguists when excluding prescribed data collection methods.
1 Introduction
From the time of the Structuralists to the present, a language documentation at minimum consists of the Boasian trilogy: a grammar, a collection of texts, and a dictionary.
The method for eliciting lexical items for a dictionary has also not changed much since
the introduction of the Swadesh list (Swadesh 1955): most frequently a single linguist,
using a Swadesh list, will elicit individual lexical items from a single speaker. As noted
by Chelliah & de Reuse (2011) the dictionary has often been ignored by field linguists
possibly due to limitations on time in the field and the linguist’s research interests. Modern linguistic field methods and language documentation handbooks, however, devote
Bryan D. Gelles. Dictionary Day: A community-driven approach to dictionary compilation. In
Jason Kandybowicz, Travis Major, Harold Torrence & Philip T. Duncan (eds.), African linguistics
on the prairie: Selected papers from the 45th Annual Conference on African Linguistics, 105–116.
Berlin: Language Science Press. DOI:10.5281/zenodo.1251720
Bryan D. Gelles
an entire chapter on doing ethical fieldwork, focusing on collaboration with the speech
community as well as encouraging linguists to give back to the speech community whenever possible, most frequently in the form of a tangible item such as a sketch grammar
or dictionary.1 During my own research, the first thing the speech community asked
for was a dictionary of the language I was documenting. The compilation of a dictionary, regardless of its exhaustiveness, is one of the most straightforward ways of giving
the community a physical manifestation of a documentation. Furthermore, Hill (2012)
recounts a tumultuous field setting where the creation of a dictionary provided positive benefits to the speech community in the form of recognition of the language and
the building of self-esteem in the community. A dictionary, thus, is not only a book of
definitions but is a cultural icon for the speech community.
The question, however, is whether the method of compilation affects the type of lexical data is collected, whether for crosslinguistic comparison or theoretical application.
If so, what methodological approaches should be used in order to strike a balance between the linguist’s (and the larger linguistic community’s) own goals and the desire to
conduct an ethical documentation. Furthermore, the linguist must consider whether the
status of the language affects what methodological approach should be taken. First I will
discuss the current methods employed in the field for dictionary compilation as well as
the implications for the wider linguistic community and the problems that come with it.
Then I will present an alternative method I used when compiling a dictionary of a highly
endangered language spoken in rural Ghana as well as discussing its implications and
drawbacks. Finally I will offer concluding remarks.2
2 Current methodology
The current methodology for eliciting lexical items for a dictionary is mostly the same
across fieldwork guides. Since Swadesh (1955), common practice has been for the linguist
to elicit individual lexical items from members of the speech community. At one end of
the extreme, Vaux et al. (2007) advocates for the lexical items to either come from a
Swadesh list or a frequency list from a related language.
I
you
we
this
that
who
what
not
all
many
one
two
big
long
small
Figure 1: Sample portion of a Swadesh list
Mosel (2004), however, notes that a predetermined list such as the Swadesh list may
present problems such as the absence of the lexical item in the speech community. For
this reason, Mosel (2004) advocates for ‘active eliciting’ whereby the linguist asks for
1 Among
2
others Bowern (2008); Chelliah & de Reuse (2011), and Vaux et al. (2007).
It should be noted the discussion itself is limited to lexical item elicitation and not specifically lexicography.
A good summary of lexicography’s own idiosyncratic problems can be found in Haviland (2006).
106
7 Dictionary Day: A community-driven approach to dictionary compilation
words related to a topic chosen by the linguist (for example, items in the home). Bowern
(2008) also suggests allowing the community members to have limited agency by having
the linguist ask them to do things like show them around the house and name items.
The linguist, in all situations, records the lexical item for the dictionary. The linguist
then must confirm the data with other speakers of the language through translation
and backtranslation to account for inconsistencies throughout the speech community
as well as to account for mistakes made on both the part of the linguist and individual
community members. The data is then compiled into a dictionary.
Another way of eliciting data is by using field guides such as those found at the Max
Planck Institute for Evolutionary Anthropology.3 These stimulus kits consist mainly of
pictures that the linguist points to, hopefully eliciting a lexical response as well as questionnaires. The linguist then records the data and uses it to compile a dictionary.
Figure 2: Example of topological relations stimuli (Bowerman & Pederson
1992).
Current language documentation methodology focuses on the audiovisual documentation of interactions with the speech community. At minimum, a microphone should
capture the audio signal of both the speaker and the linguist to not only account for the
lexical item but also the prompt (an invaluable resource when there are discrepancies
among speakers). There should also be a video recording of the interaction to account for
visual cues that may aid in spontaneous elicitations (the linguist may forget what exactly
a given stimulus was for a lexical item whereas a recording will not). Dense metadata is
compiled for each speaker and each interaction to account for not only foreseen circumstances (possible age, gender, dialect distinctions, etc.) but also for unforeseen circumstances (anything the linguist is currently unaware of about the community that may
eventually play a factor in language differences across the community). The audiovisual
component may capture those things that the linguist may either miss or misconstrue
while gathering lexical items.
3 http://www.eva.mpg.de/lingua/tools-at-lingboard/tools.php
107
Bryan D. Gelles
In all of these instances, the linguist supplies the theme if not the individual items
themselves, and (hopefully) the speech community provides a lexical item. After translating and backtranslating across multiple speakers, the linguist records what he or she
believes is the consensus across the speech community. This data is compiled into a dictionary of the language with the necessary caveats in place regarding exhaustiveness.
3 Implications
It is not a coincidence that most linguists rely on a Swadesh list for gathering lexical
items. The use of cognates to establish the genetic relationship between languages using
the Comparative Method predates the Swadesh list, but the advent of the Swadesh list
made this work much easier by codifying the list of words used for elicitation: since
linguists were using (roughly) the same set of words in the field, typologists could use
the data collected for direct comparison between languages. Also, by focusing on such
things as numbers and color terms in a language, typologists are able to compare across
multiple languages and language families relying on field linguists to gather this data
during individual documentations. It is not uncommon for typologists to contact field
linguists in order to see whether their documentations have such data necessary for
typological work, an example of the synergy between typologist and field linguist that
only a shared resource such as a Swadesh list can provide.
Stimulus materials are also used for crosslinguistic analysis by relying on the field linguist to gather very specific data the typologists and theorists cannot gather themselves
due to logistical constraints. One example of this is the Pear Story, a student-made film
that has grown to become a resource for analyzing crosslinguistic strategies for storytelling (Chafe 1980). Much like the Swadesh list, by having multiple field linguists use
the same stimulus materials, typologists and theorists can analyze a specific type of data
across languages without having to enter each speech community individually themselves, saving both time and finite resources.
4 Problems
There are several problems with these methods, however. As noted in Mosel (2004), the
Swadesh list may not line up isomorphically with the language being discussed leading
to inconsistencies: not only is it possible for an item on the Swadesh list to not be specific
or general enough for the language in question (for example, a language that does not
distinguish between the hand or the arm of a person), but it is possible that the item
does not have a correlate in the language leading to an embarrassed speech community
member who feels he or she is not up to the task at hand.4 Though the former seems like
a straightforward situation that will be easily noted by the linguist, other subtler lexical
distinctions could be lost due to strict adherence to a predetermined list. The Swadesh
4
During my own research, one community member, after not knowing a lexical item, checked every word
with a family member during the rest of the session.
108
7 Dictionary Day: A community-driven approach to dictionary compilation
List (or a frequency list) thus may inherently fail to capture the semantic boundaries of
the language while also possibly discouraging community members in the process.
The most problematic result of both using lists as well as ‘active eliciting’ is the difficulty in capturing cultural patterns due to working with individual speakers. The most
common issue facing field linguists wishing to elicit lexical items is what to do with
items that speakers disagree on. It is common in the field for one speaker to state that
the lexical item is one thing, whereas a different speaker will insist that the first speaker
has no idea what they are talking about and that the ‘real’ lexical item is something else
entirely. In many cases, this may simply be due to dialectal differences between speakers, but if the linguist is unaware of such differences, this generalization may be lost and
simply reduced to one speaker being incorrect. Furthermore, several speakers may differ from other speakers of the language. If the linguist discovers such a difference exists,
he or she may deduce a generalization exists, but if the linguist only encounters a few
members of the speech community and they all agree due to a small random sampling,
this generalization is lost. In short, by only eliciting, translating, and backtranslating one
speaker at a time, the linguist must assume the few speakers that were consulted were
prototypical of the entire speech community, a flawed assumption statistically.
Another major problem with the Swadesh list specifically is how to elicit the items
themselves. As can be seen in Figure 1 above, the Swadesh contains not only flora and
fauna (which again may not exist in field site) but also items such as personal pronouns.
Although the Swadesh list is suggested by field manuals, there is no explanation for how
to actually elicit these items. As any linguist who has ever tried to elicit personal pronouns can attest, such lexical items are tricky at best to elicit. Furthermore, without any
practical elicitation strategy to work from, each linguist creates their own method for
elicitation often having to learn by trial and error what works and potentially misconstruing the data in the process. Although the list itself is codified, the way for eliciting
it is not potentially leading to mistakes on the part of the linguist.
The linguist has also taken the majority of the agency of the documentation. The
speech community, at best, has a choice among prescribed topics and at worst must
merely translate from a list the linguist chooses. In this way the speaker is no longer a
consultant who works with the linguist to document the language but is merely an informant who does the linguist’s bidding. If the documentation is indeed a collaborative
effort (a major emphasis from an ethical standpoint), it is disconcerting that the linguist
is making decisions without the speech community’s input in regards as to what lexical items comprise the language’s dictionary. Literally, the linguist is telling the speech
community what is appropriate for a dictionary that the linguist is partially using as
justification that he or she is giving back to the community. In this power dynamic, the
linguist has all the power, and the speech community is merely a group that has the data
the linguist wants.
From a practical position, working with individual speakers is also a waste of time
and resources. There is limited time in the field, and the linguist must manage this time
wisely in order to accomplish as much possible. Working with individual speakers and
then translating and backtranslating across individual speakers uses up not only time
109
Bryan D. Gelles
but the resources the linguist is allocating for working with speakers of the community
(compensation in whatever form the linguist deems appropriate). These resources could
be used for other things that further the documentation as a whole instead of using
primarily for the gathering of lexical items.
For these reasons, the current methods for gathering lexical items are insufficient.
There is, however, another way to gather lexical items in a way that emphasizes collaboration while making differences among speakers clearer to the linguist. The question,
though, is whether too much is then lost in terms of synergistic activities with typological and theoretical linguists.
5 Dictionary Day
From December 2012 until May 2013 I began the initial documentation of Animere, a
Kwa language spoken in the rural northern Volta Region of Ghana. Previous contact
with the community produced a sociolinguistic profile as well as a short wordlist used
for comparative purposes (Ring 2006). The language is highly endangered, numbering
around thirty speakers in one isolated village. The community consisted of cocoa farmers
who work every day except for one day a week when the local market was held. On the
morning of this day, all of the speech community was invited to participate in ‘Dictionary
Day’,5 a two hour period to discuss lexical items before they went to the market. Since
the community determined it wanted a dictionary of their language, we agreed that I
would use my linguistic resources to transcribe those items they deemed appropriate for
their dictionary. They would decide on a topic for the day (or I would suggest a topic if
they were at a loss for where to begin), and I would transcribe what they told me was
appropriate for their dictionary. As this is a moribund language, it was common for the
children of the speakers to come and watch the commotion, since Dictionary Day had
a tendency to become rather lively at times due to disagreements. The dictionary that
is being compiled of this language is organized based on the topics that the community
(and sometimes myself) chose, including flaura, fauna, and traditional occupations. At
the suggestion of one speaker, their dictionary includes useful phrases in the language
as well. The dictionary, thus, is mostly their own work with the linguist performing the
role of linguistic consultant as opposed to the guider of the elicitation.
6 Methodology
As opposed to the other methods for gathering lexical items, Dictionary Day is an attempt to gather the entire speech community at one time.6 For reasons that are obvious
this is not feasible in most field situations but is fully possible when the entire speech
5 The
name and the basic idea was first suggested by Dr. Jack Martin based on his collaboration with an
American Indian speech community.
6 It should be mentioned that Bowern (2008) states in passing that working in small groups was beneficial
for collaborative reasons I will also mention for a larger group setting.
110
7 Dictionary Day: A community-driven approach to dictionary compilation
community is both small and local to the field site, and as will be seen this presents
unique benefits that cannot be gained in much larger speech communities. The speech
community is arranged in a circle, allowing each member full access to the conversation.
The linguist is also a part of this circle as both physically and symbolically an equal part
of the collaboration.
Figure 3: Dictionary Day
If the speech community meets one day a week, they are given the entire week to
think about and discuss among themselves what they would like to be a part of their
dictionary. By the time the linguist arrives on Dictionary Day, the topic will usually be
selected already by the community. If this is not the case, the linguist can suggest topics
that are appropriate to the speech community, allowing the community to determine
whether they would like to proceed with the topic suggested.
Once a topic is suggested, the community members are asked to spontaneously suggest items for the dictionary. This will only have to be done once: the community will
not need much prodding to suggest items in the future. With each topic the community
members will discuss among themselves not only the appropriateness of the lexical item
but also what forms to include in the dictionary. The linguist will then transcribe this
form and use their linguistic expertise to identify relevant information about it for the
sake of the dictionary, fulfilling their prescribed role of linguistic consultant.
111
Bryan D. Gelles
An additional list of lexical items should be kept by the linguist with the community’s
permission. In this dictionary will be all the items that were controversial, noting the
controversy surrounding the item and later, with help of the audiovisual record, what led
to the disagreement. It will be this information that will shed light on the folklinguistics
of the speech community as will be discussed further below.
All sessions should be recorded audio-visually, preferably from at least two angles if
possible to capture all the community members. The audio component will rely on microphones with wide ranges in order to capture the spontaneous speech of the community
members. For this reason microphone stands are essential: not only will the community
start to speak more spontaneously without the constant reminder of a microphone that a
linguist pointing at them would entail, but it is also impossible for a linguist to use a single microphone to capture all of the spontaneous interactions of the speech community.
Although the linguist will be transcribing the dictionary on the spot and writing dense
metadata about the session, it is these recordings that will reveal some of the missing
cultural information the speaker does not know about the language ecology of the field
site as will be explained below.
7 Implications
From an ethical standpoint, this method is ideal. The problem with the other methodologies is that they rely on the linguist to make all of the important decisions regarding
what will go into the dictionary. As discussed above, if the linguist uses a predetermined
list, the dictionary in effect becomes his or her work with the speech community only
serving as informants rather than consultants of the project. Since current ethical guidelines call for a collaborative effort, the collaboration should not only extend to working
with community members but also where possible to essentially work for them as well.
It is worth stating that the majority of a language documentation has traditionally been
to the benefit of the linguist as opposed to the speech community. This is one small way
that the community itself is able to direct the documentation of their own language.
From a purely linguistic perspective, this method also alleviates most of the problems
of the aforementioned methods. The question of how to elicit lexical items thus becomes
moot. Instead of wondering how to elicit such items from the Swadesh list as ‘louse’ or ‘I’,
the speech community will suggest items, negating any need for the linguist to invent
idiosyncratic ways to elicit lexical items. Also, the problem of speakers not knowing
a lexical item is no longer relevant as well. As Bowern (2008) notes, having multiple
speakers during a session is beneficial in that speakers will be able to prompt each other
on certain items that are little known among the speech community. This will alleviate
the pressure on the speakers to perform for the linguist and will instead merely require
the speaker to speak when comfortable, thus not endangering the linguist’s relationship
with individual speakers.
Another added benefit of this method is that disagreements among speakers are no
longer in the hands of the linguist. As mentioned above, navigating discrepancies among
speakers using a prescribed list falls on the linguist, since the linguist is meeting speakers
112
7 Dictionary Day: A community-driven approach to dictionary compilation
one at a time. Thus, if one speaker disagrees with another, the linguist must determine
which speaker’s item is suitable for the dictionary. This could cause a rift between the
linguist and those speakers’ items that were left out of the dictionary, since it is the
linguist who determines the veracity of each item. If the decision is left to the community,
this no longer becomes a problem. From a practical standpoint, it is also incumbent on the
linguist working with individual speakers to determine what constitutes a representative
sample. Field manuals mention translating and backtranslating as a way of policing data,
but they fail to mention just how many times it is required before an item is acceptable to
add to the dictionary, leaving the choice to each individual linguist. Such an unsystematic
approach could lead to idiosyncratic data, a situation often found when dealing with
older language data. This linguistic policing of data is no longer the job of the linguist but
falls onto the speech community, the group that has a better knowledge of the language
and the idiosyncrasies that come with it.
Disagreements, however, are also important for linguistic information that is normally
unavailable to a linguist working with a new speech community one member at a time.
Through disagreements among the speech community, the linguist can glean sociolinguistic information about the language. During a heated debate during Dictionary Day,
two groups formed, arguing about which lexical item was most appropriate for the language. Both sides claimed the other was wrong, and neither was willing to give any
ground. Through mediation among other members of the speech community, a form was
selected for their dictionary. My dictionary of the language, however, has both, because
the two groups that were arguing belonged to different age groups: the age-mates of one
group were arguing with the age-mates of the other. Though currently unprovable, this
suggests that there may be a generational difference linguistically that I may have not
seen if I had approached each member one at a time. During another session, the leader
of the speech community suggested an item, and everyone automatically supported the
item due to the speaker’s prestige. One speaker, however, disagreed, telling me privately
that another form was preferable. This form turned out to be an extension of a morphological pattern that I had not seen previously. Without this quiet reaction from a member
normally not vocal, I would not have seen the pattern. In this way, through various spontaneous disagreements over otherwise uninteresting lexical items, I was able to discover
both sociolinguistic data as well as a linguistic pattern I would not have been able to see
previously.
One major benefit of Dictionary Day that has thus far been assumed is the idea of
consensus among speakers. Using traditional methods, consensus is a matter of the linguist determining just how many members are necessary to constitute a representation
of the entire language. When working with a small speech community, this can be done
by speaking to each community member individually, but, as mentioned, disagreements
must be navigated somehow by the linguist. By bringing the entire community together,
however, consensus can be built among the community itself. By discussing items individually among themselves, they are literally forming a consensus for each item one
by one. Verification is done on the spot without any need to recheck most items individually. When a major dispute occurs, however, it becomes necessary to approach in-
113
Bryan D. Gelles
dividual members of the community to determine what constitutes speaker differences.
This, however, is only limited to major disputes, whereas the traditional method requires
rechecking every item. In short, actual consensus among the community can be reached
by having the entire community present at one time as opposed to choosing a number
of speakers to individually confirm lexical items.
Finally, from a practical standpoint, Dictionary Day saves both time and resources.
Instead of having to allocate the beginning of each day to checking and rechecking various lexical items speaker by speaker, the linguist can use one day a week to go over the
same amount of words while freeing up the rest of the week to work on other things.
Since each lexical item is verified at the time of its suggestion, no additional time is required, and more lexical items can be elicited quickly and efficiently. Also, whatever the
linguist deems appropriate in terms of compensation to the community will be used towards other things besides gathering lexical items, a boon to the linguist who may have
personal goals in mind in the field.
Dictionary Day thus solves the problems presented by the traditional method of gathering lexical items. Through real collaboration with the speech community, the linguist
is not only ethically interacting with the community but also doing it in a way that that
benefits his or her own research goals by freeing up additional resources. More importantly, the idiosyncrasies of the data can be worked out in a group setting without the
linguist becoming the arbitrator. The linguist may also discover language patterns that
would not be visible when speaking to only a single member of the community, a help to
the field linguist who is documenting a language that has not been analyzed previously.
8 Problems
When compared to methods that require the linguist to choose topics that speakers then
supply lexical items for, Dictionary Day is preferable in all respects. However, when compared to the use of prescribed lists or stimulus kits, Dictionary Day has a major drawback,
namely synergy among theorists, typologists, and field linguists. As previously mentioned, by using a Swadesh list, field linguists are supplying comparative linguists with
data that they themselves cannot obtain. Also, by using stimulus kits, the field linguist
is no longer supplying theorists and typologists with the same kind of crosslinguistic
data. Although language documentation is itself becoming an independent field with its
own goals, it is still preferable for documentarians to work with other linguists rather
than isolate themselves in their subfield. A common refrain among documentarians is
that it is not their job to orient their documentation around prescribed data collection
methods by theorists and typologists. It is also a common refrain among documentarians of understudied languages that their work is often ignored by those same theorists
and typologists that they themselves refuse to work with. By building a documentation
collaboratively with the speech community, the data gathered becomes idiosyncratic in
that it may not fulfill any needs of other linguists due to the random sampling of data in
the field. In this way, Dictionary Day further exacerbates this problem by not only not
limiting the data to prescribed areas of interest to other linguists but also by possibly
failing to address such areas at all.
114
7 Dictionary Day: A community-driven approach to dictionary compilation
It is, however, worth noting that many field linguists choose to use a Swadesh list
not due to any concern with other linguists’ interests but due to not contemplating an
alternative, and it is very common for fieldwork to go unnoticed regardless of the field
linguist’s intentions to the contrary. These are much larger problems than one methodology could possibly address, but it is worth mentioning those areas where the methodology fails to bridge the gap between documentarians and other linguists. For this reason,
Dictionary Day should be used in collaboration with more traditional methods. A simple
way of addressing this issue is to add Swadesh list items whenever possible to Dictionary
Day itself when the community allows. Stimulus kits could also be added, though practically it seems out of place in the context of lexical item elicitation. Whenever possible,
both traditional methods and Dictionary Day should be used side-by-side in order to not
only address the problems of the former but also to account for the problems with latter.
In this way, the documentarian can work with other linguists while not compromising
the collaborative goals of the documentation.
9 Conclusion
Dictionary Day is a way for a field linguist to work collaboratively with a speech community as a whole in situations where such a collaboration is feasible. Considering the concern of documentarians with the ethics of fieldwork, such a speech community-driven
collaboration is preferable, since it gives the agency to the community as opposed to
the linguist who has traditionally not only had all of the power but mostly uses such
power to guide the documentation in the direction of his or her own research interests.
Although direct elicitation is making a comeback (Matthewson 2004), allowing speakers
to spontaneously suggest lexical items reduces the problems of elicitation such as data
reliability. It also benefits the data collection by not only offering a different mechanism
for dealing with disputes among community members but also using such moments to
gain insights into the language itself. Consensus is thus built among the entire speech
community and not left to the linguist to determine what arbitrary number constitutes
speech community consensus. Practically, it also saves time in the field for furthering
the documentation in other ways while the linguist is in the field.
Problematically, though, Dictionary Day fails to account for linguists who need crosslinguistic data. By focusing solely on what the community chooses to do, the field linguist is not feeding more new and interesting data into the comparative, theoretical, and
typological discussion that a Swadesh list or stimulus kit would. For this reason, Dictionary Day should be used in collaboration with other methods whenever possible. The
community’s wishes must come first, but the linguist still has an obligation to the field
if he or she hopes to address such issues as the absence of understudied languages in linguistic theory. Although documentarians and other linguists sometimes have disputes
about the exhaustiveness of linguistic typology and theory, the impetus is on the documentarian to enter the discussion as well. By combining both methods, the field linguist
can find a way to bridge the divide between documentation and theory.
115
Bryan D. Gelles
Acknowledgements
This research was made possible by a grant from the Endangered Languages Documentation Programme (SG0199).
References
Bowerman, Melissa & Eric Pederson. 1992. Topological relations picture series. In
Stephen C. Levinson (ed.), Space stimuli kit 1.2, 51. Nijmegen: Max Planck Institute
for Psycholinguistics.
Bowern, Claire. 2008. Linguistic fieldwork: A practical guide. New York: Palgrave.
Chafe, Wallace L. (ed.). 1980. The pear stories: Cognitive, cultural, and linguistic aspect of
narrative production (Advances in discourse processes). Santa Barbara: Praeger.
Chelliah, Shobhana & Willem de Reuse. 2011. Handbook of descriptive linguistic fieldwork.
New York: Springer.
Haviland, John Beard. 2006. Documenting lexical knowledge. In Jost Gippert, Nikolaus
P. Himmelmann & Ulrike Mosel (eds.), Essentials of language documentation, 129–162.
Berlin: Mouton de Gruyter.
Hill, Deborah. 2012. One community’s post-conflict response to a dictionary project. Language Documentation and Conservation 6. 273–281.
Matthewson, Lisa. 2004. On the methodology of semantic fieldwork. International Journal of American Linguistics 70. 369–415.
Mosel, Ulrike. 2004. Dictionary making in endangered speech communities. In Peter
Austin (ed.), Language documentation and description, Endangered Languages Project,
39–54. London: School of Oriental & African Studies.
Ring, J. Andrew. 2006. We have no one to sing our songs: Concerns of an African elder.
Presented at the International Workshop on the Documentation and Description of
GTM Languages.
Swadesh, Morris. 1955. Towards greater accuracy in lexicostatistic dating. International
Journal of American Linguistics 21. 121–137.
Vaux, Bert, Justin Cooper & Emily Tucker. 2007. Linguistic field methods. Eugene: Wipf
& Stock.
116
Chapter 8
Language endangerment in
Southwestern Burkina: A tale of two
Tiefos
Abbie Hantgan-Sonko
Most of the thirty or so small-population languages of southwestern Burkina Faso are still
reasonably viable in spite of the spread of Jula as the dominant regional vernacular. An
unusual case is Tiefo, which is really two distinct but closely related and geographically
contiguous Gur languages. One, here dubbed Tiefo-N, was spoken in the villages of Noumoudara and Gnanfongo (Nyafogo). The other, Tiefo-D, was spoken in the nearby village
cluster of Dramandougou. Several other ethnically Tiefo villages in the zone had already
been completely Jula-ised by the mid-20th Century. Tiefo-N is moribund (a handful of ageing semi-speakers in Gnanfogo, none in Noumoudara), the villagers having gone over to
Jula. By contrast, Tiefo-D is in a relatively comfortable bilingual relationship to Jula and is
still spoken to some extent even by children, though everyone also speaks Jula. This paper
clarifies the relationship between Tiefo-N and Tiefo-D and addresses the question why the
two languages have had such different fates.
1 Tiefo
Tiefo (pronounced [čɛfɔ]) is an important ethnic group in southwestern Burkina Faso.
There are some 20 villages that still consider themselves ethnically Tiefo. The core is
constituted by the villages of Noumoudara, Gnanfogo, and Dramandougou,1 the latter
two being really clusters of several distinct physical settlements. This core is located directly on (in the case of Noumoudara) or to the east of the highway from Bobo Dioulasso
to Banfora. There are other Tiefo villages scattered around, including one to the west of
Bobo Dioulasso (on the road to Orodara) and others east and southeast of the core.2
Tiefo belongs to the large Gur language family, which dominates much of Burkina
Faso (including the large-population Mooré language of the Mossi ethnicity) and spreads
westward into parts of Ghana, Niger, Togo, Benin, and Nigeria. Manessy (1982), who
1 Alternative
spellings are Numudara, Nyafogo, and Daramandougou or Daramandugu.
village of Tiefora, east of Banfora on the road to Sideradougou and Gaouwa, is not far from Dramandougou, but in spite of its name it is apparently not Tiefo ethnically.
2 The
Abbie Hantgan-Sonko. Language endangerment in Southwestern Burkina: A tale of two Tiefos.
In Jason Kandybowicz, Travis Major, Harold Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected papers from the 45th Annual Conference on African Linguistics, 117–132.
Berlin: Language Science Press. DOI:10.5281/zenodo.1251722
Abbie Hantgan-Sonko
worked out the genetic sub-groupings within Gur, examined unpublished Tiefo data
from André Prost and concluded that Tiefo constituted its own subgroup, with no especially close relatives.
The published descriptive material on Tiefo primarily includes Kerstin Winkelmann’s
invaluable monograph (in German) on Tiefo-D Winkelmann 1998). It consists of a descriptive reference grammar (emphasising phonology and morphology) and a basic lexicon. Winkelmann was part of a German-staffed project on Gur languages and cultures
that was active in the 1990’s but has now disappeared due to retirements of senior personnel and career switches by Winkelmann and others. Her fieldwork was carried out in
Dramandougou, but she also did brief survey work (core lexicon and a little morphology)
on Tiefo-N.
Winkelmann commented that Tiefo-N, even during her fieldwork period (1990–94),
was at a much more advanced state of decline than Tiefo-D. She was able to elicit a little data from two elderly men in Noumoudara and somewhat more from semi-speakers
in Gnanfogo. The Tiefo-N lexical material was included, alongside Tiefo-D data, in her
lexicon. She calculated cognate counts for the Swadesh 100-word list between Dramandougou and either Noumoudara or Gnafongo in the 75–77 percentage range, with cognates partially disguised by sound changes and grammatical differences. She stated flatly
that Tiefo-D was not understood in either of the Tiefo-N communities.3 On the other
hand, there was good inter-comprehension between Noumoudara and Gnafongo. A reasonable conclusion is that Tiefo-D and Tiefo-N are distinct languages using normal linguistic (as opposed to political) criteria.
Given Winkelmann’s description of the dire language situation in Gnanfogo in the
early 1990’s, I was rather surprised to find some speakers in Tiefo-N in that village when
I arrived in the Bobo Dioulasso area about a decade later in 2012. In retrospect, it may
be that Winkelmann slightly underestimated the state of Tiefo-N in Gnafongo during
her brief stay there, in part because of a misunderstanding of nominal plural formation.
She stated that Gnafongo informants had difficulties producing such plurals, which a
reader could understand as implying that the language was only imperfectly remembered by a few semi-speakers. It turns out, however, that Tiefo-N pluralises many nouns
by lengthening the final vowel, i.e. singular …Cv1 becomes …Cv1v1. This corresponds
to the productive Tiefo-D plural with -r followed by a copy of the stem-final vowel, i.e.
…Cv1 becomes …Cv1-rv1. Evidently Gnafongo Tiefo-N lost the *r and the remaining identical vowels coalesced into a long vowel, a phonetically subtle pluralisation process that
could be missed during short-term fieldwork by a linguist who was not primed to look
for it.
Given the urgency of the language situation and the lack of substantial documentation
of Tiefo-N, I did some 5 months fieldwork with elderly Gnafongo speakers between August 2013 and the following January. Subsequently, Jeffrey Heath collected flora-fauna
terminology for Tiefo-N and local Jula in Gnafongo.4 In order to illustrate some of the
3 “Die in den beiden weiteren untersuchten Dörfern gesprochenen Cɛfɔ-Dialekte weichen ganz erheblich von
dem von Daramandugu ab. Weder in Nyafogo noch in Numudara ist das Daramandugu-Cɛfɔ verstehbar”
(Winkelmann 1998: 5).
4
Aminata Ouattara, a Burkina linguistics student of ethnic Tiefo origin, was also continuing fieldwork on
Tiefo-N as of early 2015.
118
8 Language endangerment in Southwestern Burkina: A tale of two Tiefos
true consequences of language contact, a greatly misunderstood phenomenon in West
Africa, I show the examples of two varieties of one moribund language. I argue that our
methodology is no longer data driven, and that because we have a certain set of ideals in
place as to what happens when one language comes into contact with another, we are
blind to the real circumstances. Instead of mourning so-called “language death” (Nettle & Romaine 2000; Price 1984), we should be celebrating the diversity of new mixed
languages which are born when speakers come into contact with one another. Through
an examination of different sociological, historical, and geographic paths, we see that
one language has become in fact two. However, without an interdisciplinary methodology that starts from the ground up, our theoretical footing will be unsound and vice
versa. In order to illustrate the differences between the presently existing Tiefo varieties,
and because there has been such little attention paid to Tiefo-N, I present an overview
and comparison of the major grammatical features of Tiefo-N and Tiefo-D. The main
phonological features are illustrated in §2 and the morphology in §3. §3.4 discusses the
differences in the pronominal (which in turn is related to the tense/aspect) systems of
the two varieties, discussed in the following section, 3.5.
Then, §5 provides an exploration of the reasons thus far provided in the literature concerning the different fates of the Tiefo villages. While geographical and sociolinguistic
reasons have been referenced in the past, the current discussion explores the historical
causes of the divergent dialects.
2 Phonology
Tiefo-N and Tiefo-D have similar consonant inventories: stops plus palatal affricates /p
b t d tʃ dʒ k ɡ kp ɡb/, nasals /m n ɲ ŋ ŋm/, fricatives /f s ɣ ʕ/, glottal /ʔ/, and nonnasal
sonorants /w l r j/. Note the distinction between the voiced pharyngeal /ʕ/ (cf. Arabic)
and glottal /ʔ/.
Table 1: Tiefo consonantal inventory.
Plosive
Nasal
Fricative
Affricate
Approximant
Trill/tap
Labial
Alveolar
pb
m ŋm
f
k͡p ɡ͡b
w
td
n
s
cj
l
r
Palatal
Velar
ɲ
(ʃ)
kɡ
ŋ
ɣ
Pharyngeal
Glottal
ʔ
ʕ
y
Absent from the consonantal inventory of both languages are several consonants reconstructed for Proto-Gur (Naden 1989): voiced implosives /ɓ ɗ ʄ/, voiced palatal stop /ɟ/,
voiced affricate /dʒ/, and labiodental fricative /v/.
119
Abbie Hantgan-Sonko
Tiefo-N and Tiefo-D likewise have similar vowel inventories, which are shared with
other languages of the zone. There are seven vowel qualities, including high /i u/, low
/a/, and two pairs of mid-height vowels, [+ATR] /e o/ and [-ATR] /ɛ ɔ/. The high and
low vowels are ATR-neutral and may combine with either type of mid-height vowel.
In Tiefo-D (Winkelmann 1998: 20, 23) but not Tiefo-N, phonemes /i u/ have optional [ATR] phonetic variants in words with a following [-ATR] mid-height vowel. Proto-Gur
is reconstructed with a ten-vowel system, including [±ATR] distinctions in high and low
as well as mid-height vowels.
i
u
e
o
ɛ
ɔ
a
Figure 1: Tiefo vocalic inventory.
Tiefo-N and Tiefo-D also have the same three tone levels. High tones are marked by
an acute accent [á], low tones by a grave accent [à]. Mid tones are written either without
an accent (Winkelmann 1998) or more explicitly with a macron [ā].
In spite of the nearly identical phonemic inventories between the two languages, many
actual pairs of Tiefo-N and Tiefo-D cognate words are disguised by phonological differences. Some examples are in Table 2, which pools data from Winkelmann (KW) and
myself (AH). Correspondences that occur in more than one set even in this small corpus
are Tiefo-D glottal stop or zero for Tiefo-N medial [ɡ], Tiefo-D [c] for Tiefo-N [s], and
Tiefo-D [d] for Tiefo-N [ʒ, j].
Table 2: Tiefo cognates.
Tiefo-D (KW)
Tiefo-N (KW)
Tiefo-N (AH)
Gloss
blaʔa ~ bla
dráⁿ
brà(ʔà)
buɔⁿ
ceʔe
cicí
cùru
dè
dɛ
báráɡà ~ báláɡà
dáraɡá
bàɡàle, bàrài
bɔʔɔⁿ, bɔɔⁿ
sereɡe
sisiu
suru
ʒàɡa, yèà
ʒɔ
bārāʔá
dárá
bàɣàʔè
būɔ ⁿ
sérííⁿ
ʃíʃíʔī
sūsúⁿ
jéjāʔā
ndɛ
‘river’
‘home’
‘hair’
‘dog’
‘skin’
‘urine’
‘millet cake’
‘sun’
‘elder brother’
120
8 Language endangerment in Southwestern Burkina: A tale of two Tiefos
3 Morphology
Morphological features found in Tiefo-N but not in Tiefo-D are a definite prefix (§3.1),
a specific set of plural suffixes (Section3.2), and an ablaut-like system of adjective-noun
agreement (§3.3).
3.1 Definite prefix
The dialect of Tiefo-N in Gnafongo has what I will call a definite prefix (but see below
for qualms about this categorisation). It has three variants depending on the dominant
vowel of the stem: [e-] before nouns with an [e] vowel in the stem, [o-] before nouns with
a back vowel [o ɔ u], and [a-] before nouns with [a] or [ɛ] vowel in the stem. Examples
are in Table 3. The stem ‘moon’ irregularly has [a-] instead of expected [e-].
Table 3: Tiefo definite.
Noun (Tiefo-N, Def-Sg)
Gloss
è-kēʔēⁿ
è-jōēⁿ
ē-sāè
ò-ŋōʕō
ō-fláɲō
ò-sīɔ̄ⁿ
ò-ɲū
à-bītɛ̄ʔɛ̀
à-fērēé
ā-kɛ́rɛ̄ ɛ̄
ā-fíyāʕā
‘spoon’
‘neck’
‘ground’
‘mosquito’
‘baobab’
‘salt’
‘water’
‘leaf’
‘moon’
‘hand’
‘field’
The definite marker is generally optional in the singular but in some cases is obligatory
in the plural. However, when the noun is followed by a quantifier or by an adjective, the
definite prefix is omitted. This suggests that the “definite” prefix functions in part to
indicate that the noun is free of modifiers.
This is more clearly the case in Tiefo-D. (Winkelmann 1998: 132) describes the Tiefo-D
prefix [e-], infrequently [o-], as obligatory in citation forms. She confirms for Tiefo-D
that it vanishes in the presence of a determiner (possessor, demonstrative).
3.2 Plural suffixes
Proto-Gur has been reconstructed as having a complex system of noun class markers in
the form of paired singular-plural combinations Naden (1989), along the lines of other
Niger-Congo families including Bantu. Many extant Gur languages still have class suffixes, and some have prefixes as well (Miehe et al. 2012).
121
Abbie Hantgan-Sonko
In addition to lengthening of the final vowel (mentioned above), a number of other
singular/plural relationships occur in Tiefo-N. Examples are shown in Table 4.
Table 4: Tiefo-N plural suffixes.
Singular
Plural
Gloss
a.
nāmi
yō nāmí
ɲō
bī
ŋmāʕa bí
ō-nāmī-jō
yō nāmí-jō
ō-ɲí-jō
bī-jō
ŋmāʕa bí-jō
‘child’
‘fruit’
‘person’
‘baby’
‘star’
b.
cɔ́mī-ī
ɲɔ́mī-ī
n̄ -cɔ́mī
ɛ̄ -ɲɔ́mī
‘bird’
‘toe’
c.
yē
jāá bɔ̄ ⁿ
yē-ʔé
jāá bɔ̄ -ɔ̄ⁿ
‘year’
‘girl’
d.
ɡbé-ēⁿ
ɡbē
‘stool’
e.
ʒɔ́wɛ̄ⁿ
ɛ́-ʒɔ́wīⁿ
‘neck’
f.
fēreʔé
fērēʔē
‘moon’
g.
dɔ̄ -jɛ̄
dɔ̄ -rɔ̄
‘man’
There are also some nouns that appear to have no singular-plural difference, such as
[búɡúnɛ ] ‘beans (variety)’, either because of recent morphological loss or because these
nouns do not lend themselves to individuation.
Winkelman reported a Tiefo-D plural /-O/ (by which she indicates an archiphoneme
representing either for [o] or [ɔ] depending on the [ATR] class of the stem), though for
animates only. This corresponds to the [-jō] (always after i) in (4a), though often not
in the same words across the Tiefo varieties. Some of the Tiefo-N glosses in (4a) are
inanimate (‘star’, ‘fruit’), but these are compounds including ‘child’ or ‘baby’, e.g. ‘treechild’ = ‘fruit’. The stem ‘man’, (4g) is a rare case where Tiefo-N has a plural [-rV] (with
copied vowel quality), the productive plural in Tiefo-D. Other Tiefo-N singular/plural
patterns (4b-f) lack known Tiefo-D matches, and are difficult to connect to reconstructed
inventories of Proto-Gur noun class markers listed by Naden (1989).
3.3 Adjectival harmony
In Tiefo-N, the final vowels of certain adjectives harmonise with the vowel of the definite
prefix of the modified noun. Consider the forms for harmonising ‘black’ in examples
(1-2) and for nonharmonising ‘big’ (2-4). The vowel quality of the prefixes on ‘house’
and ‘man’ match that of the final-vowel of ‘black’. This may reflect an archaic suffixal
122
8 Language endangerment in Southwestern Burkina: A tale of two Tiefos
agreement pattern, creating a construction of the type *[CLASS-noun adjective-CLASS].
Synchronically it could be described as a terminal ablaut (i.e. mutation of the final vowel
into another quality). There is no similar mutation of the adjective ‘big’, which has an
invariant shape in (2-4).
(1)
wà- wūʕú jōb-á
def hut black
‘the black house’
(2) ò- dɔ̀ɛ̀ jób-ō
def man black
‘the black man’
(3)
à- wūʕú sã̄ɡbānāʔà
def hut big
‘the big house’
(4) ò- dɔ̀ɛ̀ sã̄gbānāʔà
def man big
‘the big man’
3.4 Pronouns
The subject personal pronouns of Tiefo-N are those in Table 5. The singular but not
plural forms vary depending on the aspect (perfective/imperfective) of the clause (imperfective includes progressive). The basic Tiefo-D forms (Winkelmann 1998: 140) are
shown for comparison; specifically imperfective (‘present’) and negative Teifo-D combinations are omitted. Tiefo-D distinguishes animacy in the 3sg, and also has uses the
distant demonstrative [bó] as a discourse-anaphoric 3sg pronoun.
Table 5: Tiefo pronouns.
1sg
1pl
2sg
3pl
3sg
3pl
Tiefo-N Imperfective
Perfective
Tiefo-D
ɲí
é
mì
nā
kā ō
ɲí
ān
é
m
nā
nō
ān
no
ʔejuò
mo
buò
ʔɔ̄ⁿ (anim), ʔà (inan), bó (anaph) ʔò
no
For Tiefo-N, 1sg subject is exemplified examples (5-6), 1pl in (7-8).
123
Abbie Hantgan-Sonko
(5)
ɲí
wɔʕɔ̀ bè kṹ
1sg.ipfv prog come today
‘I am coming today.’
(6) n bàʔ jànā
1sg come yesterday
‘I came yesterday.’
(7)
é wɔʕɔ̀ bè kṹ
1pl prog come today
‘We are coming today.’
(8)
é bàʔ jànā
1pl come yesterday
‘We came yesterday.’
Unlike Tiefo-D, Tiefo-N does not currently distinguish animacy or anaphoricity (e.g.
reflexives) in the 3sg pronoun. This might be due to recent grammatical simplification,
and the occasional use of Jula 3sg pronouns shows that language contact has impacted
the pronominal system.
3.5 Verbal aspectual inflection
Verbal aspectual morphology in Tiefo-N is more intricate than nominal or pronominal
morphology. The main opposition is between imperfective and perfective (sometimes
called ‘continuous’ and ‘neutral’, respectively).
In one verb class, the imperfective is unsuffixed while the perfective is marked by a
low- or mid-toned suffix -ra ~-la Table 6. It can be nasalised to -na, see ‘arrive’ (Table 6,
row (d)).
Table 6: Tiefo aspectual affixation.
a.
b.
c.
d.
e.
Imperfective
Perfective
Gloss
jē
jè
bīɛ́
dã̄
dīò
jé-rā
jē-rà
bīɛ̄ -rà
dā-nà
dīō-là
‘enter’
‘walk’
‘farm’
‘arrive’
‘sell’
Several other verbs show ablaut-like vocalic mutations, in some cases along with other
internal changes or affixes. Two multiply attested patterns are vowel to [a] (row (a) in
Table 7) and [a] to [e/ɛ] (row (b) in Table 7). Mutation types attested once are in (row (c)
in Table 7).
124
8 Language endangerment in Southwestern Burkina: A tale of two Tiefos
Table 7: Tiefo aspectual mutation.
a.
b.
c.
Imperfective
Perfective
Gloss
sè
bè
bē
díʔī
dōʔò
dōrōʕò
ɲānā
náʔā
dárāà
bārá
jāʕà
ɲ-à
bó
sá
bāʔ
b-là
díā
dāà
dárāʕā
ɲéné
nɛ́nɛ̄
dɛ̄ rɛ̀ɛ̀
bērẽ̀
jɛ̄ ɡɛ̀
ɲ-ū
bwɛ̄
leave/go
come
tire
eat
plant
buy
stop/stand
wash (clothing)
rip
sweep
break
drink
tie
An important difference between the two Tiefo varieties is that Tiefo-N has a preverbal morpheme wɔ ʕɔ that marks progressive aspect. No similar preverbal progressive or
imperfective morpheme is reported for Tiefo-D. It is possible, however, that the Tiefo-N
form is archaic, reflecting a proto-form *bo ‘be’ Manessy (1982).
4 Influence from Jula
The data in 6 consist of verbs which are suffixed with [-rV] or allomorphs [l]~[n] in an
aspectual form known as ‘neutral’ or perfective. The suffix may be a borrowing from Jula
since the perfective suffix in Jula is [-ra] with allomorph [-la]. An example illustrating
the [-rV/lV] suffix in Gnanfongo Tiefo is the verb ‘hide’, borrowed directly from Jula as
[dūɡū], ‘hidden’ [dūɡū-là]. Many of the verbs in this category are probable borrowings
from Jula, even though a neutral suffix [-da/ra/ta] is attested in other Gur languages.
However, according to Naden (1989), most verbal markers are treated as particles rather
than affixes in other Gur languages. The most widely marked inflectional category in Gur
languages is expressed through a contrast between the continuous (imperfective) and a
form described as ‘neutral’. Therefore, the neutral suffix in Tiefo is likely related to the
particle found in other Gur languages, but possibly has been reanalysed in Gnanfongo
Tiefo as a perfective suffix on Jula borrowings.
According to lexical comparisons between Tiefo and other Gur languages by Manessy
(1982), there is but a mere 28 out of 435 correspondence, 20 percent. With sample correspondences shown in Table 8, between the data gathered by the author from Gnanfongo
Manessy’s (1982) from Dramandougou and surrounding Gur languages, we do see, however limited, some strong evidence for a a related source.
125
Abbie Hantgan-Sonko
Table 8: Correspondences between Tiefo and Gur languages (Manessy 1982:
146)
Tiefo (AH) Tiefo (GM) Viemo Doɣose Gan
pĩĩ́ ́
pini
pinyɔ pĩĩ́ ś e
kāʕà
sã́ã́
kaʔa
sãã
kaasɔ kaase
saasi -sãã
ɲēréē
fērēʕé
nāfāʕɔ́
ɲã̄
yāá
sāʕè
bēʕé
káʕá ɲīn
ɡ͡bã̄ ã́
ɲinde
fereɡi
donu
ɲã
ya
sari
bẽ
kaane
bɔ̃, baa
ɲɛɲɛ
ferɡe
doni
Lobi Dyan
Kulango Loron Gloss
pĩĩ́ ́
kasa
-sãã
ɲeɲa
filiki
doŋko
ɲeɲa
-sãã,
-sãzi
ɲuɡo
piniɡu, ‘excrement’
pininyu
‘meat’
-sã
‘three’
ɲã
yɛrɛ
siru
bənə
kannɔ
baawɔ
bana
sáákɔ̀
bẽ
‘breast’
‘moon’
‘slave’
‘give’
‘woman’
‘earth’
‘wilderness’
‘tooth’
‘sheep’
Manessy gives three hypotheses for how non-Gur roots are found in Tiefo: Tiefo
should be placed within a separate branch of Gur, certain words are borrowed from
an unknown Gur language, or the source of the borrowing is non-Gur, possibly Mande.
If language contact from Jula were the dividing factor, one would expect there to be clear
borrowings from Jula into Tiefo. If the Jula language is an influence, it would be apparent
in the lexicon.
Among plant names, we find evidence for a sustained symbiosis between Jula and even
Tiefo-N. For example, Heath recently recorded flora-fauna terms in Gnanfogo, both in
Tiefo-N and in the local Jula. Quite a few of these terms are phrasal, and the Tiefo-N and
local Jula often share the phrasing. Some plant names are in Table 9.
These correspondences, though limited to natural species terms, are indicative of a
broader pattern of calquing, the effect of which is develop a local Tiefo-ized Jula. Outside the core Tiefo area, this must have the same general sociolinguistic function of
marking speakers as Tiefo, as we observe with familiar ethnically-tinged English varieties (Yinglish, Spanglish, and the like).
While we do see some evidence of borrowing from Jula in both dialects of Tiefo in
the Table 10, according to comparisons between my data and Winkelmann’s shown in
Table 11, most are like the second table, with 87 out of 185 core lexical items do not bear
any resemblance between the two dialects, nor to Jula (based on my knowledge of Jula).
The evidence presented from the lexicon shows that Jula has not influenced either
variety of Tiefo to the point that one would expect if the majority language were to be
blamed for the loss of the minority language. Considering the long term contact of Jula
with Tiefo, one would expect more of an influence on the lexicon than what is found.
Further, the lexical differences between the two dialects, for the most part, cannot be
126
8 Language endangerment in Southwestern Burkina: A tale of two Tiefos
Table 9: Tiefo plant names.
Tiefo
Jula
Identification
Literal
sòy-pûŋ
lɛ̀-bííⁿ
‘pig-herb’
pô:ŋ-sà:ⁿ-wi
bí:ŋ-ŋwání-tígi
bàwáⁿ-sāní
sàmà-ŋwánì
cò:-kú:ⁿ
sùlà-fíⁿsáⁿ
Acanthospermum
hispidum
Amaranthus
spinosus
Asparagus
africanus
Cola cordifolia
bàwáⁿ-dùrté
sàmá-tìsékàà-bé
nàfɔ́ɣɔ́ⁿ-bàkó-èllè-wí
blákè-póróŋ
nɔ̀ɣɔ̀sì-dúy
jààtìgì-fáɣá
sándéⁿ-wɔ̀rɔ̀sɔ́
nɔ̀ɣɔ̀sì-kúú
kàⁿkóóⁿ-tòè
sòfàlì-túló
bɛ̌ yⁿ-jùsúⁿ
kòŋó-jèsé
sóⁿ-bàⁿflà-glá-yò
sò-tìgí-bàⁿflà-bɔ
sèsèré-dúy
bàsàⁿ-kúù
blákɛ́-flɔ̄
sándéⁿ-sìrà-yírí
wámbíí-ʃìnàà
fárátá-dɛ́bɛ́
sìsàɣà-dúrúŋ-tè-pô:ŋ
kámmélé-kóróbóo
Combretum
nigricans
Ficus thonningii
Heeria insignis
Heliotropium
indicum
Leptadenia
hastate
Securidaca
longepedunculata
Senegalia
macrostachya
Stachytarpheta
indica
Sterculia setigera
Uapaca
togolensis
clumpy grass sp.
’herb-thornowner’
‘elephant-thorn’
‘monkeycashew.apple’
‘elephant can’t
knock it down’
‘host-kill’
‘rain-sickle’
‘chameleon-tail’
‘donkey-ear’
‘outback-wire’
‘horseman-hattake.off’
‘agama-tail’
‘hare’s baobab
fruit’
‘orphan-mat’
‘young.man-testgrass’
attributed to influence from Jula on either end of the dialect spectrum. The cause of the
divergences within Tiefo and within Gur must have been triggered by another source,
but it remains unknown.
127
Abbie Hantgan-Sonko
Table 10: Potential borrowings from Jula into Tiefo.
Tiefo Dramandougou
Tiefo Gnanfongo
Jula
Gloss
ɡūɡlīká
blanà(-nɔ)
nāklɔ̄
po-jenɔ̄ , poka
ɲã
jūwɛ́ʔaɛ́
dɔ̀̃
jɔ̄ , jɔ̄ -rɔ̄
wòrò
kērē kīté
míɔ̄ nɔ̄
mīɔ̄ nɔ̄
ō-dòsō
sɔ̄ ʕɔ́
ɡānāʕà
náfāʕō
bíkā
dōʕōbíyō
kɔtɛ
mali
malo
donso
so
ɡalaji
jɔn
jo
wòro
‘snail’
‘hippopotamus’
‘rice’
‘hunter’
‘horse’
‘indigo’
‘slave’
‘fetish’
‘kola nuts’
Table 11: Cross-dialectal lexical non-concordance not due to Jula influence.
Tiefo Dramandougou
Tiefo Gnanfongo
Jula
Gloss
sú
sɛ̀ɡɛ̀
sɔ́ʔɔ́, sɛ́ʔɛ́
sīɡlòʔó -ro
ɡ͡bɛ bà
pūʔō, poʔo
dɛ̀, bɛ-tɔʕɔ
diɛ̀
baʕa
jūwɛ́ʔaɛ́
sàk͡pè
dúrú
dúwī
dūwõ̀
fáʕláī
fīyāá
fíyāʕā
fīyáʕā
fíyɔ̀
fíʕī
ɡānāʕà
kā kɔ̃́
ɲinan
dimi
cin
suruku
lana
kunɡo
foro
buɡu
tɔmɔ
ɡalaji
fali
‘mouse’
‘hurt’
‘sting’
‘hyena’
‘take’
‘wilderness’
‘field’
‘multiply’
‘pick up’
‘indigo’
‘donkey’
5 Why different fates?
The preceding discussion demonstrates that Tiefo-N and Tiefo-D are two distinct, though
closely related languages. Why have they suffered such different fates?
Isolation? Perhaps Dramandougou (Tiefo-D) is more isolated than Gnanfongo and
Noumoudara (Tiefo-N). Well, it is true that Noumoudara is directly on the Bobo Diolasso
to Banfora highway, and this may have been the coup de grace factor for Tiefo-N in that
village. But Gnanfongo and Dramadougou are both located in the same lowlands area
southeast of a long escarpment that cuts them off from the highway. Both are reached
from the highway with some difficulty, by 4x4 or a motorcycle, either by taking a southern route that avoids the cliffs or by winding one’s way down a circuitous descent in
a relatively benign part of the escarpment between Noumoudara and Gnanfongo. Gov-
128
8 Language endangerment in Southwestern Burkina: A tale of two Tiefos
ernment institutions (schools, clinics) are present in Gnanfongo and Dramandougou to
about the same extent. They are equally “isolated”.
Perhaps a vigorously expanding regional language had a more direct line of sight on
Gnafongo than on Dramandougou due to some geographical quirk? The two candidates
for “killer” languages (Nettle & Romaine 2000; Price 1984) would be French and Jula. Indeed it was once feared that French and English would give the same scorched earth
treatment to African languages as English has given to the indigenous languages of Australia and North America. This has now been broadly debunked by Batibo (2005) and
Mufwene (2009). In West Africa, even in sophisticated and heavily Gallicized coastal
megalopolises like Dakar and Abidjan, French has developed symbiotic relationships
with other languages rather than eliminating them, and new synthetic formations such
as Nouchi and Urban Wolof are emerging. In villages far from the coast like Gnanfongo
and Dramandougou, French is a minor factor in the sociolinguistic equation. Naden (1989:
141) makes the point that southwestern Burkina has historically been a “backwater” relatively unaffected by the outside world, from the late medieval Saharan trade routes to
the present.
Jula is another matter. Southwestern Burkina is a linguistic mosaic of ancient Gur
languages (Tiefo, Lobi, Viemo, Dogose, Turka, and others) with interspersed Mande
languages like Bobo and Zuungo that date to the Mande expansion of the late Middle
Ages. The Bambara-Jula-Mandinke dialect group, which is also Mande genetically, has
become the linguistic juggernaut throughout southern Mali (Bamako, Segou), southwestern Burkina, and northern Cote d’Ivoire. Its spread in Burkina was spearheaded by merchants who made it into the lingua franca in markets and then in urban concentrations.
The name of the biggest city southwestern Burkina, Bobo Dioulasso (i.e. Bobo-Jula-So),
attests to the coexistence of Jula with other indigenous languages. If there is a killer
language in the area, it is clearly Jula, not French.
However, there is no obvious geographical reason why Jula should have targeted TiefoN for extinction any more than Tiefo-D. Jula is the dominant interethnic vernacular in
the entire region, extending deeply into neighbouring Cote d’Ivoire. If Dramandougou
were more isolated than Gnanfongo, Jula might have had a more powerful foothold in
the latter. But Dramandougou is no more isolated than Gnanfongo. Jula is spoken at least
as second language by everyone in Dramandougou as well as Gnanfongo.
What about strategic self-interest as an explanation? An SIL-sponsored survey of the
local situation does state that “Most Tiefo have abandoned their language in favour of
Jula … presumably as a result of a perceived social advantage to be gained by using Jula”
(Berthelette & Berthelette 2001: 5). But self-interest should be just as pertinent to Tiefo-D
as to Tiefo-N. As Showalter (2008) states in his survey of the languages of Burkina Faso,
only two communities in the entire country replaced their languages with Jula, one being
Tiefo-N and Lüpke & Storch counter such simplistic reasoning: “there is no evidence of
which we are aware where the shift to another language (as opposed to maintaining it
as a language in a multilingual repertoire) has yielded real socio-economic advantages”
(Lüpke & Storch 2013: 286).
129
Abbie Hantgan-Sonko
What about differential “prestige” as an explanation? Aside from the elusiveness of
this concept,5 the fact is that Tiefo ethnic pride is if anything stronger in the Tiefo-N
than Tiefo-D area, and perhaps stronger there than in the other small-population ethnicities in the area between the proud, larger-population Bobo and Lobi. The background to
this is that the Tiefo tribe was a feared military power until the turn of the 20th Century.
To this day there is a Tiefo “chef de guerre” in Noumoudara, distinct from the regular political chief. He commands no battalions, but he does supervise a small military museum
dedicated to the memory of an early chief named Amoro Ouattara. In this museum, visitors get guided tours recounting the great battles of the past and demonstrating (gently)
the uses of the traditional weapons, shields, and torture equipment that are on display.
It is not large, but it is more than the other small-population ethnicities in the area have.
In Africa and elsewhere, language coexistence (multilingualism) is the norm, not the
exception. There is no zero-sum fight to the death among languages. Again (Mufwene
2009: 76): “Such a practice of language alternation is traditional to Africa and has sustained multilingualism, so much so that it takes a natural disaster to force whole villages
to move and find themselves in situations where they have to shift to the host population’s language.”
The cataclysmic event that accelerated the decline of Tiefo was the military victory
of the Jula leader Samori Touré over the Tiefo, followed by the slaughter of many Tiefo
people in 1897. This is cited as the key event in the demise of the language by Hébert
(1958), Le Moal (1980: 31), and Winkelmann (1998: 2). It is likely that the Tiefo-N villages
who commanded the Tiefo forces were the principal victims.
Dramandougou, on the periphery and not centrally involved in military activity, appears to have already reached an accommodation with the Jula, resulting in a less confrontational relation, at the time of those hostilities. For that reason it was spared the
brunt of the reprisals.
6 Conclusion
Despite the fact that there are only five speakers in the village of Gnanfongo, all in their
70’s and 80’s, the dialect of Tiefo differs from the neighbouring village, particularly in
the lexicon. The differences between the two dialects of Tiefo cannot be due to Jula alone.
In fact then, language contact, in addition to not “killing” a language, may not have as
much influence as we think.
Languages, differing from the metaphors we like to invoke of species, rarely simply die
out without a trace, rather, they converge into and diverge from one another. Speakers
do not suddenly one day wake up and decide it will be advantageous to being speaking
another language. The history of many countries in Africa and the world is volatile,
with environmental and political factors influencing language to a greater degree than
we may account for. The example of the Tiefo serves not only to illustrate that we are
5
In the early days of American sociolinguistics, the core idea was that lower middle-class individuals sought
to emulate the speech of the highest local socioeconomic class. But the data eventually forced recognition
of, first, a kind of prestige in the lower echelons, and then another kind of prestige in the middle.
130
8 Language endangerment in Southwestern Burkina: A tale of two Tiefos
missing pieces in the history of the people, but also that we are ill equipped to gather
those pieces given the framework we have been using.
Although the cause of the loss of the Tiefo language can with a fair amount of certainly
be attributed to Samori Toure and his army of invaders, beyond that, the discrepancies
between the existing Tiefo dialects which cannot be attributed to Jula remains a mystery.
In summary, Tiefo shares some features of geographically neighbouring Gur languages
but does not fit into any known branch of Gur. Further, the variety of Tiefo that remains
in the lives of the five elderly speakers in Gnanfongo differs significantly from the more
robust version of the language spoken in neighbouring Dramandougou.
Acknowledgements
This research is conducted as part of the project “Investigating the interaction of tone and
syntax in the Bangime and the Dogon languages of Mali and Burkina Faso”, funded by
BCS-1263150 (2013–16), PI Jeffrey Heath. I am grateful to Jeffrey Heath, Friederike Lüpke,
Marieke Martin, and Sophie Salffner for their contributions and support throughout the
writing of this paper. I would also like to thank the two anonymous referees for their
helpful comments.
References
Batibo, Herman M. 2005. Language decline and death in Africa: Causes, consequences and
challenges. Clevedon: Multilingual Matters.
Berthelette, John & Carol Berthelette. 2001. Sociolinguistic survey report for the Tíefo language. Tech. rep. SIL International.
Hébert, Jean R. P. 1958. Une page d’histoire voltaïque: Amoro, chef des Tiefo. Bulletin de
l’I.F.A.N B 20. 377–405.
Le Moal, Guy. 1980. Nature et fonction des masques.
Lüpke, Friederike & Anne Storch. 2013. Repertoires and choices in African languages.
Berlin: Mouton De Gruyter.
Manessy, Gabriel. 1982. Materiaux linguistiques pour servir à l’histoire des populations
du sud-ouest de la Haute Volta. Sprache und Geschichte in Afrika 4. 95–164.
Miehe, Gudrun, Ulrich Kleinewillinghöfer, Manfred von Roncador & Kerstin Winkelmann. 2012. Overview of noun classes in Gur (II)(revised and enlarged version). In
Gudrun Miehe, Brigitte Reineke & Kerstin Winkelmann (eds.), Noun class systems in
Gur languages, vol. 2, 5–37. Köln: Rüdiger Köppe.
Mufwene, Salikoko S. 2009. What Africa can contribute to understanding language vitality, endangerment, and loss. In Matthias Brenzinger & Anne-Maria Fehn (eds.), Proceedings of the 6th World Congress of African Linguistics, Cologne, 17-21 August 2009,
69–80. Cologne: Rüdiger Köppe Verlag.
131
Abbie Hantgan-Sonko
Naden, Anthony J. 1989. Gur. In John Bendor-Samuel & Rhonda L. Hartell (eds.), The
Niger-Congo languages: A classification and description of Africa’s largest language family, 140–168. New York: University Press of America.
Nettle, Daniel & Suzanne Romaine. 2000. Vanishing voices: The extinction of the world’s
languages. Oxford: Oxford University Press.
Price, Glanville. 1984. The language of Britain. London: Edward Arnold.
Showalter, Stuart D. 2008. Un profil du bilinguisme en dioula au sud-ouest du Burkina Faso.
Technical report, SIL International.
Winkelmann, Kerstin. 1998. Die Sprache der Cefo von Daramandugu (Burkina Faso). Sonderforschungsbereich 268 an der JW Goethe-Universität.
132
Part III
Morphology and phonology
Chapter 9
Consonant substitution in child
language (Ikwere)
Roseline I. C. Alerechi
University of Port Harcourt
The Ikwere language is spoken in four out of the twenty-three Local Government Areas
(LGAs) of Rivers State of Nigeria, namely, Port Harcourt, Obio/Akpor, Emohua and Ikwerre
LGAs. Like Kana, Kalabari and Ekpeye, it is one of the major languages of Rivers State of
Nigeria used in broadcasting in the electronic media. The Ikwere language is classified as an
Igboid language of the West Benue-Congo family of the Niger-Congo phylum of languages
(Williamson 1988: 67, 71, Williamson & Blench 2000: 31). This paper treats consonant substitution in the speech of the Ikwere child. It demonstrates that children use of a language can
contribute to the divergent nature of that language as they always strive for simplification
of the target language. Using simple descriptive method of data analysis, the paper identifies
the various substitutions of consonant sounds, which characterize the Ikwere children’s utterances. It stresses that the substitutions are regular and rule governed and hence implies
the operation of some phonological processes. Some of the processes are strengthening and
weakening of consonants, loss of suction of labial implosives causing them to become labial
plosives, devoicing of voiced consonants, etc. While some of these processes are identical
with the adult language, others are peculiar to children, demonstrating the relationships
between the phonological processes in both forms of speech. It is worthy of note that highlighting the relationships and differences will make for effective communication between
children and adults.
1 Introduction
The Ikwere language is spoken in four out of the twenty-three Local Government Areas
(LGAs) of Rivers State of Nigeria, namely, Port Harcourt, Obio/Akpor, Emohua and Ikwerre LGAs. Like Kana, Kalabari and Ekpeye, it is one of the major languages of Rivers
State used in broadcasting in the electronic media (Alerechi 2007a: 1). Williamson (1988:
67, 71) classifies Ikwere as one of the Igboid group of languages as well as Igbo, Ekpeye, ogba, Echie, to mention but a few. Williamson & Blench (2000: 31) locate Igboid
under the node of West Benue-Congo family of Niger-Congo phylum of languages. The
Roseline I. C. Alerechi. Consonant substitution in child language (Ikwere). In Jason Kandybowicz, Travis Major, Harold Torrence & Philip T. Duncan (eds.), African linguistics on the prairie:
Selected papers from the 45th Annual Conference on African Linguistics, 135–155. Berlin: Language
Science Press. DOI:10.5281/zenodo.1251724
Roseline I. C. Alerechi
Ikwere language comprises twenty-four divergent dialects, which are mutually intelligible. It is yet to develop a standard dialect. However, there are published works such as
Donwa-Ifode & Ekwulo’s (1987) Ikwere Orthography, Tẹsitament Iikne (a translated New
Testament Bible), and some recent scholarly works in the language.
Some of the works like Williamson (1980), Donwa-Ifode & Faraclas (2001), among
others, observe different realization of some phonological segments in Ikwere. In fact,
Alerechi (2007a) specifically identified some phonological processes responsible for the
different realizations of segments, which may have contributed to the divergent nature of
the Ikwere language. Some of such processes are loss of suction of labial implosives causing them to become labial plosives, the spirantization (weakening) of labial and alveolar
plosives to labial and alveolar fricatives, respectively, the voicing of alveolar fricatives,
etc.
It is interesting to note that these studies are focused on the adult use of the Ikwere
language to the neglect of the area concerning child language. This serves as a motivation
for this paper. Given that the general trend for children is to change the sounds of the
language in an attempt to use them, this paper is aimed at identifying such changes
and consequently the phonological processes characteristic of Ikwere children aged 3 to
4 years. Following the assertion of Fromkin et al. (2003a: 358) that early phonological
rules generally reflect natural phonological processes that occur in the adult (target)
language, this paper further investigates if the phonological processes in child language
are identical with those of the adult (target language).
This paper focuses on O̩ deegnu (Odgn), E̩ mowha (Emwh), Akpo, Aluu and Omuanwa
(Omnw) dialects of Ikwere, an Igboid language spoken in Rivers State of Nigeria. Even
though the sound substitution in child language involves both consonants and vowels,
this paper specifically focuses on the substitution of consonants of the target language
for those characteristic of the child’s language.
1.1 Literature review
This section gives a brief review of literature in child language acquisition. It specifically
treats the phonological development, phonological processes and outlining the target
sounds.
1.2 Phonological development
Communication is a natural phenomenon to every human being. Thus, to enable children
to communicate with others in their environment, they need to acquire the language.
O’Grady et al. (2011: 361) state that the ability of children to produce speech sounds
begins to emerge at around six months with the onset of babbling. Babbling enables
the children to experiment with and begin to gain control over their vocal apparatus. It
increases in frequency until the age of about twelve months, when the children begin to
produce their first understandable words.
Scholars have investigated the order of acquisition of sounds by the children and some
observed that among the earlier sounds are the back velar sounds [k]and [g], and front
136
9 Consonant substitution in child language (Ikwere)
vowels like [a], [i] and [e]. Others recognize the bilabials [m], [p] and the alveolar sounds
[n] and [d] demonstrated in such sequences as ma, pa, di (Bolinger 1975: 283; Labarba
1981: 344; Ojukwu & Alerechi 2011: 69). There is a contrary view that children acquire
velar consonants before the bilabials and alveolars (Anthony et al. 1971: 45). In line with
this view, Alerechi & Awala (2012: 257) observe that Ekpeye children below age three
replace the velar plosives [k] or [g] with the alveolar plosives [t] or [d], respectively.
This implies that despite the similarity observed in the order in which children acquire
speech sounds, individual differences still abound. In fact, each child develops his own
systematic way of producing adult forms within his limited scope of sound sequences
(Menn 1992: 813).
In spite of a good deal of variation observed from one child to the other in terms of
the order of mastering sounds in production and perception, the general tendencies as
outlined by O’Grady et al. (2011: 362) seem to exist. Based on the manner of articulation
(stricture), stops tend to be acquired before other consonants. In terms of place of articulation, labials are often acquired first followed (with some variation) by alveolars, velars,
alveopalatals.
Realizing that sounds do not exist in isolation but in sequences to form morphemes or
words, such sequences comprise vowel and consonant. Thus, vowel and consonant occur
in a sequence to make up syllable structure and children tend to simplify the syllable or
word structure of the target language. Children structures are mainly CV, CVCV (Akpan
2004: 25). This implies that in the acquisition of the adult speech by the children certain
phonological processes are in operation.
1.3 Phonological processes
Phonological processes are those changes which segments undergo that result in the various phonetic realizations of the underlying phonological segments (Yul-Ifode 1999: 144)
Children adopt certain phonological processes to attain to the adult sounds. According to
Akpan (2004: 26), phonological processes in children are short-cut processes that operate
on the child’s speech in his attempt to attain the adult target. As the child’s chronological age increases, the phonological processes decrease to conform to the phonological
system of the language. Akpan (2004: 27) further notes three major classifications of
the phonological processes: substitution, assimilatory and syllable structure processes.
In addition to these major classifications, Yul-Ifode (2003: 2) recognizes dissimilation
(intervocalic consonant devoicing) as a fourth major process. Of the four major phonological processes identified in the literature, the present paper focuses on consonant
substitution in children speech.
Substitution is the systematic replacement of one linguistic feature for another or one
phoneme for another that the child finds easier to articulate (Fromkin et al. 2003b: 357;
Akmajian et al. 2008: 491; O’Grady et al. 2011: 365. O’Grady et al. (2011: 365) identified
common substitution processes to include stopping, the replacement of a fricative by a
corresponding stop; fronting, the moving forward of a place of articulation; gliding, the
replacement of a liquid by a glide; and denasalization, the replacement of a nasal stop by a
137
Roseline I. C. Alerechi
non-nasal counterpart. Scholars like Akmajian et al. (2008), Akpan (2008), David (2009),
Akpan (2010), Alerechi & Ojukwu (2010) and Ojukwu & Alerechi (2011) have equally
identified various forms of sound substitution in child language in different languages.
The present paper treats the consonant substitution in Ikwere.
1.4 The target sounds
There are nine phonetic oral vowels [i ɪ e ɛ a o ɔ u ʊ], and eight phonetic nasalized vowels
[ı̇̃ ɪ̃ ẽ ã õ ɔ̃ ũ ʊ̃] in Ikwere (Donwa-Ifode & Ekwulo 1987: 42–43; Alerechi 2007a). They are
summarized in the vowel charts on Figure 1 and Figure 2 respectively.
i
e
ɪ
ʊ
[ɛ]
u
o
ɔ
a
Figure 1: The Ikwere phonetic oral vowels.
ı̇̃
ẽ
ɪ̃
ʊ̃
ũ
õ
ɔ̃
ã
Figure 2: The Ikwere phonetic nasalized vowel.
Contrary to Donwa-Ifode & Ekwulo (1987: 42–43), which see [e] and [ɛ] as the allophonic variants of the phoneme /e/ in Ikwere, Alerechi (2007a: 65) observed that the
vowel [ɛ] contrasts with other vowels in some dialects and is an allophonic variant of /e/
in some other dialects. It is also noted that beside the distinctive nasal vowels recorded
in the language, vowels in the environment of nasal consonant may or may not be nasalized.
On the other hand, Ikwere records thirty-one phonetic consonants as shown in Table 1.
There is no long consonant in the language.
Of the thirty-one consonants recorded in the language twenty-eight of them are phonemic. Following Alerechi (2007a: 98), the number of the phonemic consonants in each of
the dialects, however, varies. It ranges from twenty-six to twenty-eight.
All the sounds in parentheses in Table 2 occur in Ọgkr. Whereas one or more of these
sounds are allophone(s) in some dialects, and do not exist in others. Alerechi (2007a: 99)
138
9 Consonant substitution in child language (Ikwere)
Table 1: The phonetic consonants of Ikwere (adapted from Donwa-Ifode & Ekwulo 1987).
Manner of
articulation
Place of articulation
Labial
Alveolar
m
Nasal
Plosives
p
b
Implosives
ƥ
ɓ
t
f
v
Velar
ɲ
ŋ
d
k
Affricatives
Fricatives
Palatal
n
s
z
Tap
tʃ
dʒ
ʃ
ʒ
Labialized velar
Labial-velar
Glottal
ŋʷ
ɡ
kʷ
ɰ
hʷ
ɡʷ
ʔ
h
ɾ
Central
approximants
j
w
l
Lateral
approximant
Table 2: The phonemic consonants of Ikwere.
Manner of
articulation
Place of articulation
Labial
Nasal
Alveolar
m
Plosives
(p)
(b)
Implosives
ƥ
ɓ
Tap
t
ɲ
ŋ
d
k
tʃ
(f)
(v)
s
Labialized velar
Labial-velar
Glottal
ŋʷ
ɡ
kʷ
ɰ
hʷ
ɡʷ
dʒ
z
h
ɾ
Central
approximants
Lateral
approximant
Velar
n
Affricatives
Fricatives
Palatal
j
w
l
139
Roseline I. C. Alerechi
gives clear picture of the occurrence of [p b f v] in the dialects of Ikwere as shown in
Table 3 - Table 7.
Table 3 shows that /p b f v/ are phonemic in Ọgkr.
Table 3: Ọgkr.
p
f
b
v
Table 4 records /p b v/ as phonemes in the above dialects.
Table 4: Ozha, Ọmnw, Ubma, Akpb, Egbd, Elle, Omdg, Ubmn Omrl and Apni.
p
–
b
v
Table 5 illustrates that /b f v/ are phonemic in Ẹmwh.
Table 5: Ẹmwh.
–
f
b
v
The dialects in Table 6 record only /p b/ as phonemes.
Table 6: Akpọ , Obio, Alụu, Igwr Omgw, Iskp, Ipo and Omdm.
p
–
b
–
Table 7 is the reverse of Table 6 as only /f v/ are phonemic in the four dialects.
Table 7: Rmkp, Rndl, Ọdgn and Ib/Ob.
–
f
140
–
v
9 Consonant substitution in child language (Ikwere)
The foregoing Table 3 - Table 7 demonstrate the occurrence the consonants /p b f v/
in the various dialects of Ikwere. The present study, therefore, intends to find out among
other things, if children from the dialects without any of these sounds would manifest
such in course of acquiring the language.
1.5 Methodology
The wordlist used in collecting data from the subjects was drawn from the wordlist collected by Alerechi (2007b) for treating labial variation in Ikwere. It contains seventy-two
words of everyday life obtainable in the environment of the subjects. The words contain
different sounds of the language and are made up of monosyllabic, disyllabic and polysyllabic structures, giving such structures as V, CV, VCV, VCVCV, etc. The data were
collected from each of the subjects by imitation and object pointing methods at their
residence. The visit to each subject’s residence during the period of data collection was
about two to three times in order to elicit the accurate forms of the subject’s speech. The
subjects’ speech forms were recorded manually and finally transcribed for analysis. The
study adopts a descriptive approach in analyzing the data. It focuses on identifying and
analyzing the substitution patterns and processes of consonants in the speech of the Ikwere children. The data was also analyzed using SPSS. Descriptive statistics was carried
out to describe the performance of the subjects. The different occurrence of the consonants produced by the adult and subject in each dialect, were converted to quantitative
data which was presented as percentages using Bar Charts.
1.6 The subjects
The subjects consulted during the period of data collection are seven; however, five of
them were selected for the analysis because a comparison of the speech forms of two
subjects from the same dialect area showed a replication of the other. The five subjects
comprise three female and two male and fall within the age range of 3 to 4 years (3, 3, 3
½, 3 ½ , 4). They were selected from O̩ deegnu, Emowha, Akpo, Aluu and Omuanwa so
as to investigate if in the course of sound change, a child from a particular area would
manifest forms typical of those of other area(s) or not. Table 8 summarizes details of the
subjects.
Table 8: Data on the subjects.
Subject
1. VN
2. MA
3. EE
4. IN
5. GW
Gender
Age
Village
Dialect
Female
3 years
Rumuodogo
Odeegnu
Male
3 years
Elibarada
Emowha
Female
3 ½ years
Rumuolumini
Akpo
Female
3 ½ years
Omuokiri
Aluu
Male
4 years
Ubordu
Omuanwa
141
Roseline I. C. Alerechi
1.7 Consonant substitution
There are twenty-eight phonemic consonants /m n ɲ ŋ ŋʷ p b t d k g kʷ gʷ ƥ ƃ ʧ ʤ f v s
z ɰ hʷ h r j w l/in Ikwere. Some of these consonants are replaced with some others in
child language. The pattern of substitution reflects those involving different states of the
glottis, places of articulation and manners of articulation. The pattern reflecting different
states of the glottis sometimes overlap with those of places of articulation. I, therefore,
present the various substitutions based on manner of articulation.
1.7.1 Substitution according to manners of articulation
The different manners of articulation observed in the data involve the plosives, fricatives,
affricates, implosives, approximants, etc.
1.7.2 Substitution of plosive with plosive
The substitutions here reflect those involving states of the glottis or places of articulation. Hyman (1975: 16) observed that the general tendencies in child language include
the learning of voiceless stops before voiced stops. This phenomenon is identified in this
paper in the utterances of children above the age of 3 to 4. Thus, where the target language records the voiced stops [b] or [d], the tendency is for the children above age 3
to 4 to replace them with their voiceless counterparts [p] or [t], respectively. Table 9 is,
therefore, strong evidence that Ikwere children are not left out in first acquiring voiceless consonants and subsequently revert to the target forms. The table shows the Odgn
subject replacing the voiceless velar plosive [k] with the voiceless alveolar plosive [t] of
the target.
Table 9: d → t: b → p, k → t.
Odeegnu
Emowha
Akpo
Target
VN 3yrs
Target
MA 3½yrs
Target
m̀ƥòró ákâ
àƃà
zɪ
àkítì
m̀pòtó átà
àbà
ʤɪ
àtítì
m̀ƥòró ákâ
àƃà
zì
àkɪdɪ
m̀pèrè-áká
àbà
zì
àkɪdɪ
m̀ƥèré-ákā
àbã̀
zì
àkídì
EE 3½yrs
àbà
Aluu
Omuanwa
Target
IN 3½yrs
Target
GW 4yrs
ìsíní ákâ
àbà
zɪ
àkísì
ìpèní áꜜkâ
àbà
zɪ
àkítì
ḿƥèré ákâ
àbà
dɪ
ákɪdɪ
ípèlé áꜜkâ
àpà
tè
àkʷà
Gloss
‘elbow’
‘jaw’
‘is’
Beans
(brown)
There is also the tendency of the Ikwere subjects replacing complex articulated (labialized) sounds with those of simple articulation (single segments). This phenomenon is
predominant in the speech form of the Odgn child, even though children from other dialect areas manifest traces of this phenomenon. Consider the data in Table 10. Note that
the data include the simplifying of labialized fricative and nasal.
1.7.3 Substitution of plosive with fricative
The pattern of substitution treated in this section reflects those involving [v] with [b],
and vice versa, and that of [s] with [t]. While children from Emwh and Akpo show
preference for [b] instead of [v] of the target language, they, conversely, replace [v] with
142
9 Consonant substitution in child language (Ikwere)
Table 10: kw → k; kw → t; gw → g; hw → w, ŋw → m.
Odeegnu
Emowha
Akpo
Aluu
Omuanwa
Target
VN 3yrs
Target
MA 3½yrs
Target
EE 3½yrs
Target
IN 3½yrs
Target
GW 4yrs
vékʷū
εkʷã́
εkʷã̂
ɛŋʷʊ
ɛkʷà
ɛꜜŋʷâ
εhĩ̂
ʤɪ
ɔʧɪ
vétù
εtá
àtâ
εmʊ
àtà
ámà
εjɪ
ʤɪ
ɔʧɪ
békʷǔ
εkʷã́
εkʷâ
ɛŋʷʊ
ɛkʷà
ɔꜜŋʷâ
εhĩ̂
ʤɪɔʧɪ
békʷǔ
áŋkʷá
εkʷâ
εmʊ
àkʷà
εŋʷâ
àjî
ʤɪ
ɔʧɪ
békʷǔ
àkʷã́
àkʷá
àŋʷʊ
àkʷà
ɔꜜŋʷá
εhĩ ́
ʤɪ
ɔʧɪ
békʷǔ
àkã́
àká
àŋʷʊ
àkʷà
ɔꜜmá
àhɪ
ʤɪ
ɔʧɪ
békʷǔ
àkʷã́
àkʷâ
àŋʷʊ
àkʷà
ɔꜜŋʷâ
àhʷʊ
gʷʊ
ɔkʷʊ
békʷǔ
ŋʹká
áꜜkʷâ
áŋʷʊ
àkʷà
ɔꜜŋʷâ
àwʊ
gɔ
ɔkʊ
békʷù
ákʷá
àkʷâ
áŋʷʊ
àkʷà
ɔꜜŋʷâ
àhʷʊ
gʷʊ
ɔkʷʊ
békù
ákʷá
àkʷâ
áwʊ
àkʷà
ɔꜜŋʷâ
àwʊ
gʷʊ
ɔkʷʊ
Gloss
Greet
Cry
Egg
Death
Bush fowl
Moon
Body
Given (name)
Leg
[b] as demonstrated in Table 11. The data further show the children in the choice of [t]
for [s] of Aluu. The substitution in this section agrees with the observation of Crystal
(1997: 242) that the replacement of fricatives with stops is one of the possible trends for
children in language acquisition. It is worthy of note that this substitution seem peculiar
to children that are above age 3 to 4.
Table 11: s → t; v →b; b → v.
Odeegnu
Emowha
Akpo
Aluu
Omuanwa
Target
VN 3yrs
Target
MA 3½yrs
Target
EE 3½yrs
Target
IN 3½yrs
Target
GW 4yrs
tṍóvírízí
èvùlù
dívjà
tó
m̀vɔ
ɔvɔʧɪ
òvèʤè
àkítì
tɔóvílìʤì
àvùlù
ʤɔ
m̀vɔ
ɔvɔʧɪ
òvèʤì
àtítì
tòr̃ṹ
óbírízí
èvùlù
díbjà
tó
m̀vɔ
ɔvɔʧɪ
òbèʤè
àkɪdɪ
tòr̃ṹ
óbíjíz
èvùnù
dívjà
tó
m̀bú
ɔbɔʧɪ
èvèʤè
àkɪdɪ
sṍ
óbírízí
èvùlù
díbjà
só
m̀bɔ
ɔbɔʧɪ
òbòʤò
àkídì
ʧṍ
óbílízí
èbùlù
m̀bɔ
ɔbɔʧɪ
òbòjò
-
só
óbírízí
èbùlù
díbjà
só
ḿꜜbɔ
ɔbɔʧɪ
òbèʤè
àkísì
só
óvírízí
èbùrù
díbjà
só
ḿꜜbɔ
ɔbɔʧɪ
òbòʤè
àkítì
tò
óbúrúzù
m̀fùlù
díbjà
tó
ḿꜜvɔ
ɔbɔʧɪ
òbèʤè
ákɪdɪ
tòr̃ṹ
óbúlúsù
èvùlù
díbjà
tó
ḿꜜfɔ
ɔbɔʧɪ
òbòʤò
àkwà
Gloss
‘follow’
‘sympathy’
‘ram’
‘doctor’
‘grow’
‘comb (in)’
‘day’
‘mudskipper’
Beans
(brown)
1.7.4 Substitution of fricative with fricative
In addition to substituting voiced stops with their voiceless counterparts, Table 12 further proves that the replacement of voiced consonants with their voiceless counterparts
extends to the fricatives. The substitution is, however, predominant in the speech of an
Omnw child of 4years old than those of Odgn of 3years and Akpo of 31 /2 yrs as the data
demonstrate. Thus, the children replace [v] and [z] of the target language with [f] and
[v], respectively. Our data show that the changes occur both word-initially and wordmedially indicating that there is no conditioning factor for the change. The data further
demonstrate the tendency of the child from Omnw replacing some of the vowels in initial position with a syllabic nasal. This additional peculiarity observed in the speech of
the Omnw child, further strengthens the claim of the presence of a slight speech problem
143
Roseline I. C. Alerechi
in this child’s language. This, however, requires further investigation in other to confirm
our claim.
Table 12: v → f, z → s.
Odeegnu
Emowha
Akpo
Aluu
Omuanwa
Target
VN 3yrs
Target
MA 3½yrs
Target
EE 3½yrs
Target
IN 3½yrs
Target
GW 4yrs
sã́
úꜜsû
sṹ
ɔsʊ̂
ɔꜜsʊ̂
zʊ
zɔ
ézè
ɔɲì
èvùlù
ɔzà
vã̀
sã́
úʤù
ʧṹ
ɔʧʊ
ɔʧʊ
ʤʊ
ʤɔ
ɪvǔ
àvùlù
áʤà
fǎ
sã́
úꜜsû
sṹ
ɔsʊ̂
ɔsũ̀
zʊ
zɔ
ézè
ɪvù
ɪvù
èvùlù
ɔzʊzà
vã̀
sá
ńꜜsũ̂
sṹ
àsʊ̂
ɔsũ̀
zʊ
zɔ
ézè
ɪvú
ɪvù
èvùnù
àzʊzà
vã̀
zã́
iꜜzṹ
Zṹ
àzʊ́
ázʊ̀
zʊ
zɔ
ézè
ɪvù
ɪvù
èvùlù
ɔzʊzà
bã̀-
ɪʧu
zú
àzʊ́
áʧʊ
sʊ
ʤɔ
zã́
íꜜzṹ
zṹ
àzʊ̂
ázʊ̀
zʊ
zɔ
ézè
ɪbû
ɪbù
èbùlù
ɔzìzà
bã́
zá
íꜜʤû
ʤú
àʤû
áʤʊ
ʤʊ
zɔ
éʤè
ɪꜜbû
ɪbù
èbùrù
ɔʤìzà
bǎ
zã́
úꜜzṹ
zṹ
àzʊ̂
ázʊ̀
zʊ
zɔ
ézè
ívû
ɪvù
m̀fùlù
ɔzìzà
bã̀
sá
ńꜜsũ̂
sú
àsʊ̂
ásʊ̀
sʊ
sɔ
ésè
ḿfû
ḿfù
èvùlù
ɔsìsà
bà
ɪvú
ɪvù
èbùlù
ɔsìsà
bǎ
Gloss
‘imitate’
‘corpse’
‘steal’
‘back’
‘fish’
‘buy’
‘step on’
‘king’
‘load’
‘fat’
‘ram’
‘broom’
‘enter’
1.7.5 Substitution of fricative with affricate
There is also the substitution of the alveolar fricatives [s] and [z] of the adult utterance
with the palato-alveolar affricates [ʧ]and [ʤ], respectively, in children pronunciation.
Thus, where Odgn, Akpo and Aluu adult articulate [s] or [z], the choice for the children
is [ʧ] or [ʤ], respectively, indicating the affrication of these fricatives. Occasionally, the
children substitute [s] for [ʤ] or [z] for [ʧ] as demonstrated in Table 13. The substitution
Table 13: Fricative versus affricate: s → ʧ; z → ʧ, z → ʤ.
Odeegnu
Emowha
Akpo
Aluu
Omuanwa
Target
VN 3yrs
Target
MA 3½yrs
Target
EE 3½yrs
Target
IN 3½yrs
Target
GW 4yrs
úꜜsû
sṹ
ɔsʊ̂
ɔꜜsʊ̂
zʊ
zɔ
óꜜsũ̂
sʊ
úʤù
ʧṹ
ɔʧʊ
ɔʧʊ
ʤʊ
ʤɔ
óʧù
ʧʊ
úꜜsû
sṹ
ɔsʊ̂
ɔsũ̀
zʊ
zɔ
óꜜsũ̂
sʊ
ńꜜsũ̂
sṹ
àsʊ̂
ɔsũ̀
zʊ
zɔ
ósṹ
sʊ
iꜜzṹ
Zṹ
àzʊ́
ázʊ̀
zʊ
zɔ
óꜜsṹ
sʊ
ɪʧu
zú
àzʊ́
áʧʊ
sʊ
ʤɔ
óꜜsú
ʧʊ
íꜜzṹ
zṹ
àzʊ̂
ázʊ̀
zʊ
zɔ
éꜜsû
sʊ
íꜜʤû
ʤú
àʤû
áʤʊ
ʤʊ
zɔ
éꜜʧû
ʧʊ
úꜜzṹ
zṹ
àzʊ̂
ázʊ̀
zʊ
zɔ
éꜜsũ̂
sʊ
ńꜜsũ̂
sú
àsʊ̂
ásʊ̀
sʊ
sɔ
ńꜜsũ̂
sʊ́
sʊ́
ɔsʊ
sɔ
ʧʊ́
ɔʧʊ
ʧɔ
sʊ́
ɔsʊ
sɔ
sʊ́
ɔsʊ́
sɔ
sʊ́
ɔsʊ
sɔ
ʧʊ́
ɔʧʊ
ʧɔ
sṹ
ɔsʊ
sɔ
sú
ɔsʊ
sɔ
sʊ́
ɔsʊ
sɔ
sʊ́
ɔsʊ
sɔ
tṍzɪ
ǹzí
óvírízí
tɔ
ʤɪ
ǹʤì
óvílìʤì
tṍ
zì
ǹzí
óbírízí
tòzì
sṍ
zì
ǹzí
óbírízí
ʧṍ
ǹzí
óbílízí
sṍ
zɪ
ǹzí
óbírízí
só
zɪ
ǹʤí
óvírízí
sõ̀
dɪ
ǹdí
óbúrúzù
tòtè
ìdí
óbúlúsù
144
óbíjíz
Gloss
‘corpse’
‘steal’
‘back’
‘fish’
‘buy’
‘step on’
‘millipede’
‘pound
(yam)’
‘wash’
‘bat’
Forbid or
respect
Follow
‘is’
‘husband’
‘sympathy’
9 Consonant substitution in child language (Ikwere)
of alveolar fricatives with palato-alveolar affricates is restricted to the child of 3years,
whereas the replacement of the voiced alveolar fricative with the voiceless counterpart
seems peculiar to the child of 4. This phenomenon, though, geographically determined
is observed in the adult speech, the impression of this paper is that it may be a case of
speech impediment in the utterance of this 4 year old child.
1.7.6 Substitution of fricative with approximant/null
The majority of the children have not acquired the glottal fricative [h]. The data in Table 14 demonstrate that they either delete it wherever it occurs in the target form or
replace it with [j]. See Table 14.
Table 14: h → j, h →∅.
Odeegnu
Emowha
Akpo
Target
VN 3yrs
Target
MA 3½yrs
Target
EE 3½yrs
hã́
ã́
-
-
ƥě
pě
úhjé
εhĩ̂
újeá
εjɪ
úhjé
εhĩ̂
újé
àjî
ɪhjé
εhĩ ́
ɪjé
àhɪ
Aluu
Target
ɪhjé
àhʷʊ
Omuanwa
IN 3½yrs
Target
GW 4yrs
-
ƥé
pè
ɪjé
àwʊ
úhjí
àhʷʊ
újé
àwʊ
Gloss
Peel
(orange)
Red
Body
1.7.7 Substitution of tap with lateral or nasal
The tendency is also recorded of Ikwere children to use the alveolar lateral approximant
instead of tap or nasal. Thus, where the adult use the alveolar[r], the children show preference for the alveolar lateral [l] or the alveolar nasal [n]. This substitution is characteristic
of children that cut across ages 3 to 4 as shown in Table 15.
Table 15: r →l, l → r.
Odeegnu
Emowha
Akpo
Aluu
Omuanwa
Target
VN 3yrs
Target
MA 3½yrs
Target
EE 3½yrs
Target
IN 3½yrs
Target
GW 4yrs
ɛƥã́rã́
óvírízí
m̀ƥòró ákâ
ùrì
rí
ŋʷɔ
ɔꜜlô
ɔpálá
óvílìʤì
m̀pòtó átà
ùlì
ɔrɔ
ɔƥã́rã́
óbírízí
m̀ƥòró ákâ
ùrì
rí
ŋʷɔ
óꜜlô
ɔpálá
óbíjíz
m̀pèrè-áká
ùlì
rí
ŋʷṍ
óꜜrô
ɔƥã́rã́
óbírízí
m̀ƥèré-ákā
Ìrì
rí
ŋʷɔ
óꜜló
ɔpárá
óbílízí
ìlì
ɔꜜrɔ
ɔƥã́rã́
óbírízí
ìsíní ákâ
ìrì
rí
rí
éꜜlô
ɔpáná
óvírízí
ìpèní áꜜkâ
ùlì
lí
wɔ
áꜜrɔ
ɔƥã́rã́
óbúrúzù
ḿƥèré ákâ
ùrì
rí
rí
éꜜlô
ɔpálá
óbúlúsù
ípèlé áꜜkâ
ùlì
lí
lí/lílí
éꜜlô
Gloss
‘first son’
‘sympathy’
‘elbow’
‘indigo’
‘eat’
‘drink’
‘antelope’
1.7.8 Substitution of implosive with plosive
The replacement of the labial implosive [ƥ] of the target language with the labial plosive
[p] serves as another trend in the speech of Ikwere children. Thus, where the choice in
the target language is [ƥ], the children use [p] as demonstrated in Table 16. The data show
that the substitution of [ƥ] with [p] cuts across ages 3 to 4. A similar trend is observed
145
Roseline I. C. Alerechi
with the voiced counterparts [ƃ] and [b]. This implies that the acquisition of [ƥ] and [ƃ]
is a later development in the language of children.
Table 16: ƥ → p, ƃ → b.
Odeegnu
Emowha
Akpo
Aluu
Omuanwa
Target
VN 3yrs
Target
MA 3½yrs
Target
EE 3½yrs
Target
IN 3½yrs
Target
GW 4yrs
ɛƥã́rã́
m̀ƥòró ákâ
ƥʊ́
ƥó
ɔpálá
m̀pòtó átà
-
ɔƥã́rã́
m̀ƥòró ákâ
ƥʊ́
ƥṍ
ɔpálá
m̀pèrè-áká
pé
pó
ɔƥã́rã́
m̀ƥèré-ákā
ƥʊ́
ƥṍ
ɔpárá
Pó
ɔƥã́rã́
ìsíní ákâ
ƥʊ́
ƥõ̌
ɔpáná
ìpèní áꜜkâ
pé
pǒ
ɔƥã́rã́
ḿƥèré ákâ
ƥʊ́
ƥṍ
ɔpálá
ípèlé áꜜkâ
pè
pó
hã́
ã́
-
-
ƥě
pě
-
-
ƥé
pè
àƃà
àbà
vṍ
àƃà
vú
àbà
ƥṍ
àbã̀
Pó
àbà
bṍ
àbà
bó
àbà
bṍ
àbà
bó
àpà
Gloss
First son
Elbow
Scrape
Pack
waste
Peel
(orange)
Accuse
Jaw
The data in this paper demonstrate that sound substitution in child language also
involves the vowels. From the data, however, the replacement of sounds involving vowel
are not as recurrent as those of the consonants. This agrees with previous studies that
vowels are acquired earlier by children than consonants. By the age of 3 years most
vowels sounds would have been established hence, no need for much substitution. Tonal
substitution in the language is not significant in children speech as sown in most of the
data.
1.8 Substitution processes
The various substitution patterns observed in this paper give additional evidence of the
simplification of adult (target) language by children. It is observed that as children develop, the substituted sounds are dropped to conform to the adult forms when they have
gained greater articulatory control. Recalling that the substitutions are not haphazard
but rule governed, an interesting question would be what rules do children impose to
simplify the adult language? In other words, what are the phonological processes operating to relate the child utterances with the target forms? Considering the divergent
nature of the Ikwere language, Alerechi (2007a) identified a number of phonological processes relating one form of speech of a particular geographical location with the form of
the others, one of which is the reflex of the proto-form, while others are likely innovations. Thus this section does not only identify the phonological processes in operation,
but also draws attention to the processes that are identical with those of the adults as in
the speech of different geographical areas and those that are typical of child language.
The following subsections discuss the phonological processes observed in this paper.
1.9 Strengthening and spirantization (weakening)
Bearing in mind the replacement of plosives with fricatives and vice versa, I observed
the processes of strengthening and spirantization (weakening), respectively, in Table 11.
Alerechi (2007a: 262) earlier observed the spirantization (weakening) of [b] of some di-
146
9 Consonant substitution in child language (Ikwere)
alects to [v] in some other dialects but not the strengthening of [v] to [b]. In child language, however, there is an addition of the phonological process of strengthening of [v]
of the adult form to [b] showing that the process reflecting the change of [v] to [b] is
typical of children in Ikwere, whereas that involving the change of [b] to [v] is identical
with the adult speech.
1.10 Loss of suction and a shift to plosive
Concerning the substitution of [ƥ] with [p], we note a loss of suction of labial implosive
of the adult language and a shift to labial plosive in child language. Alerechi (2007a: 260)
recognized this phonological process in Ikwere. Here children from the dialect areas with
the preponderant use of [ƥ] show preference for [p] of some other dialect areas. This
change does not only reveal that the phonological process in child language is identical
with the adult language, it also proves that phonological substitution characteristic of
child language is rule governed.
1.11 (Successive) affrication (and (de)voicing)
For the change of fricatives with affricates, two phonological processes are in operation.
One is the affrication of [s] to [ʧ] and [z] to [ʤ] depending on voicing. Another is the
successive affrication and voicing or devoicing of fricatives as in [s] changing to [ʤ]
and [z] changing to [ʧ]. Even though Alerechi (2007a: 261) observed affrication in adult
language, this affrication is different as it reflects [s] and [ʧ] or [z] and [ʤ] and not [t],
[ʧ], [ʃ] and [s] as in the adult language.
1.12 De-labialization
De-labialization is a process whereby the feature of lip rounding on the primary stricture
is lost. In Table 10, it is observed that all the subjects replaced the main stricture in either
one or more of these changes [kʷ→ k; kʷ→ t; gʷ→ g; hʷ→ w, ŋʷ→ m]. Even though a
child may have produced any of the labialized sounds accurately, labialized sounds are
among the difficult consonants for the children to acquire since all the subjects manifested de-labialization process.
1.13 Gliding
Gliding in child phonology is a process whereby any consonant is realized as a glide
(Yul-Ifode 2008: 255). In addition to producing accurately [j] as in the target form, the
Odeegnu and Emowha subjects produce [h] as [j] intervocalically, indicating that [j] is
among the early sounds in child language acquisition.
147
Roseline I. C. Alerechi
1.14 Denasalization
Even though this paper focuses on consonants, we observed a preponderance loss of
nasalization of nasalized vowels in child language in our data especially in Table 16.
Thus the phonological process of denasalization of certain vowel segments of the target
language is evident of child language. Though Alerechi (2007a: 249) observed denasalization in the adult language, it is not as predominant as that in child language, showing
that loss of nasalization is one of the ways children actually simplify their pronunciation.
In fact nasalizing a vowel segment requires extra energy or force and more natural for
children to neglect it than the adults. This explains why we record a preponderance loss
of nasalization in children speech forms more than those of the adults.
2 Interpretations
The foregoing substitutions of segments in the language of Ikwere children are evidence
of simplification of items of the target pronunciation of the Ikwere consonants. In an
attempt to articulate the sounds of the language, Ikwere children, like other children,
change the sounds when trying to attain the target form thus, resulting in imperfect rendering of some of the target sounds. The imperfect representations of the adult sounds
generate ambiguous forms and this contributes significantly to the communication gap
existing between children and adults. While the children understand the adults, but
find it difficult to communicate effectively, the adults, on the other hand, have achieved
greater articulatory control of the target language, but cannot grasp fully the intentions
of the children. This could create serious problems such as frustration on the part of the
adults, particularly, the impatient ones and dissatisfaction, resentment and most likely,
seclusion on the part of the children if the gap is not bridged as the adult may not reach
their needs. However, it is the parents that understand their children better than any
other person.
The various phonological processes observed in §1.5, which demonstrate that the processes in both adults and children’s forms of speech are to a great extent identical with
slight differences also has implication. The similarities indicate that the gap in communication existing between adults and children could be bridgedparticularly if the adults are
conversant with the forms of the varieties spoken in other geographical locations. For
the forms peculiar to children, an exposure to what should be expected could facilitate
communication and reduce communication gap to the barest minimum.
2.1 Performance scores of consonants
Table 17 represents the phonetic consonants of Ikwere observed in the data. It shows the
articulatory activity of each subject. The total occurrence of each consonant is obtained
by counting the number of occurrences of the sound in the words elicited from both
the target and the subjects. These were converted to quantitative data and presented as
percentages using bar charts. The bar charts displaying the performance of the children
148
9 Consonant substitution in child language (Ikwere)
from Odeegnu, Emowha, Akpo, Aluu and Omuanwa are represented in Figure 3 to Figure 7. It should be noted that twenty-three (23) consonants were observed in the data
collected. Due to dialectal difference all the consonants did not manifested in the target
of all the dialects. Thus Odeegnu records 17; Emowha and Akpo have 19 consonants each,
while Aluu and Omuanwa used 21 and 23, respectively.
2.2 Consonants in Odeegnu
Table 17 and Figure 3 show that the Odeegnu subject aged 3 years scored 100% in the
production of [ʧ] and [ʤ]. It could be that the number of these sounds is limited or that
the child has mastered them completely. The subject has acquired [v f t l d j] to a reasonable extent but not completely. The table and figure also show that the subject could
not produce [ŋʷg kʷ ƥ z r] and had difficulty in pronouncing [k]. The total performance
of the subject in the production of the target consonants is 43.2%, which is below the
average.
2.3 Consonants in Emowha
From Table 17 and Figure 4, it is observed that the Emowha male subject of 3 ½ years
produced [t d k g ʧ ʤ f s z j] accurately and above the average in [kʷ v ŋʷ]. He, however,
scored zero percent in the production of [ƥ ƃ h]. This subject seems to have acquired most
of target sounds as against those that are yet to be included in his inventory. Table 17
puts the total performance of the Emowha subject in the production of target consonants
at 75.6%.
2.4 Consonants in Akpo
From Table 17 and Figure 5, the Akpo female subject of 3 ½ years scored 100% in the
production of [p b t d k g ʧ f j], above average in [ʤ v kʷ] and has attained average score
in [ŋʷ h l]. On the other hand, she scored zero percent in the production of [ƥ] and less
than average in [r s z]. Table 17 also shows that the subject has acquired 64.3% in the
production all the target sounds.
2.5 Consonants in Aluu
Table 17 and Figure 6 show the Aluu male subject of 3 ½ years with 100% score in the
production of the consonants [ŋʷ p t d k g ƃ ʧ ʤ j], 93.8% in [b] and above the average in
[s kʷ]. However, the subject scored zero percent in the production of [gʷ ƥ h hʷ w], and
14.3% and 25% performance scores in [r] and [l], respectively. The total percentage for
the production of all the target consonants is 64.4%, indicating above average mastery
of the target sounds.
149
Sound
Odeegnu
𝑇
ŋʷ
p
b
t
d
k
g
kʷ
gʷ
ƥ
ƃ
ʧ
ʤ
f
v
s
z
h
hʷ
r
j
w
l
Total
3
−
−
4
3
3
1
4
−
2
−
2
4
5
10
11
6
3
−
6
3
−
4
74
Emowha
1
%
3
2
1
0
0
75
66.7
33.3
0
0
0
0
2
4
4
9
2
0
0
100
100
80
90
18.2
0
0
0
2
0
66.7
3
75
0
32
0
43.2
𝑇
3
−
4
4
4
4
2
4
−
5
2
4
4
4
10
11
8
2
−
7
4
−
4
90
2
2
Akpo
%
66.7
2
4
4
4
2
3
50
100
100
100
100
75
0
0
4
4
4
7
11
8
0
0
0
100
100
100
70
100
100
0
3
4
42.9
100
2
50
68
75.6
𝑇
2
1
9
2
2
1
1
5
−
3
−
4
4
1
4
7
10
2
−
4
4
−
4
70
3
Aluu
%
1
1
9
2
2
1
1
3
50
100
100
100
100
100
100
60
0
0
4
3
1
3
2
4
1
100
75
100
75
28.6
40
50
1
4
25
100
2
50
45
64.3
𝑇
2
3
16
2
3
3
3
5
1
4
1
1
2
−
−
9
13
1
1
7
5
1
4
87
4
Omuanwa
%
2
3
15
2
3
3
3
3
0
0
1
1
2
100
100
93.8
100
100
100
100
60
0
0
100
100
100
6
5
0
0
1
5
0
1
66.7
38.5
0
0
14.3
100
0
25
56
64.4
𝑇
2
2
12
2
6
4
2
7
1
5
1
1
2
2
4
8
11
1
1
8
5
1
4
92
5
1
2
11
2
4
3
2
6
1
1
0
1
2
2
0
6
0
0
0
0
5
0
3
52
%
50
100
91.7
100
66.7
75
100
85.7
100
20
0
100
100
100
0
75
0
0
0
0
100
0
75
56.5
Roseline I. C. Alerechi
150
Table 17: Frequency of occurrence for consonants.
9 Consonant substitution in child language (Ikwere)
100 100
100
Percentage
75
80
66.67
50
90
66.67
33.33
18.18
0
0
0
C
75
0
0
0
0
0
C3 C4 C5 C6 C7 C9 CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS9 S1
S3
Consonants in Odeegnu
c
c1 c2 c3 c4 c5 c6 c7 c8 c9 cs cs1 cs2 cs3 cs4 cs5 cs6 cs7 cs8 cs9 s1 s2 s3
ŋʷ p
b
t
d
k
g
kʷ gʷ ƥ
ƃ
ʧ
ʤ f
v
s
z
h
hʷ r
j
w l
Figure 3: The Odeegnu subject performance in consonants.
100100100100
100
75
66.67
Percentage
100100100
100
100100
70
50
50
0 0
0
50
42.86
0
C C2 C3 C4 C5 C6 C7 C9 CS CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS9 S1 S3
Consonants in Emowha
c
c1 c2 c3 c4 c5 c6 c7 c8 c9 cs cs1 cs2 cs3 cs4 cs5 cs6 cs7 cs8 cs9 s1 s2 s3
ŋʷ p
b
t
d
k
g
kʷ gʷ ƥ
ƃ
ʧ
ʤ f
v
s
z
h
hʷ r
j
w l
Figure 4: The Emowha subject performance in consonants.
151
Roseline I. C. Alerechi
Percentage
100
100100100100100100
100
50
75
60
50
100
100
75
40
28.57
50
50
25
0
0
C C1 C2 C3 C4 C5 C6 C7 C9 CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS9 S1 S3
Consonants in Akpo
c
c1 c2 c3 c4 c5 c6 c7 c8 c9 cs cs1 cs2 cs3 cs4 cs5 cs6 cs7 cs8 cs9 s1 s2 s3
ŋʷ p
b
t
d
k
g
kʷ gʷ ƥ
ƃ
ʧ
ʤ f
v
s
z
h
hʷ r
j
w l
Figure 5: The Akpo subject performance in consonants.
Percentage
100
100100100
100100 100100100100
93.75
100
71.43
60
55.56
50
25
0 0
0 0
0 0
0
C C1 C2 C3 C4 C5 C6 C7 C8 C9 CS CS1CS2CS5CS6CS7CS8CS9 S1 S2 S3
Consonants in Aluu
c
c1 c2 c3 c4 c5 c6 c7 c8 c9 cs cs1 cs2 cs3 cs4 cs5 cs6 cs7 cs8 cs9 s1 s2 s3
ŋʷ p
b
t
d
k
g
kʷ gʷ ƥ
ƃ
ʧ
ʤ f
v
s
z
h
Figure 6: The Aluu subject performance in consonants.
152
hʷ r
j
w l
9 Consonant substitution in child language (Ikwere)
2.6 Consonants in Omuanwa
The Omuanwa male subject of 4 years scored 100% in the production of [p t g gʷ ʧ ʤ f j];
91.7% and 85.7% in [b] and [kʷ], respectively, and 75% in [k s l]. For the difficult sounds,
the subject obtained zero percent score for [ƃ v z h hʷ r w] and 20% for [ƥ]. The total
performance of this subject in the production of all the consonants is 56.5% of the target.
These are shown in Table 17 and Figure 7.
100 100
91.67
100
100 100
Percentage
75
66.67
100100100
85.71
100
75
75
50
50
20
0
0
0
0 0 0 0
0
C C1 C2 C3 C4 C5 C6 C7 C8 C9 CS CS1CS2CS3CS4CS5CS6CS7CS8CS9 S1 S2 S3
Consonants in Omuanwa
c
c1 c2 c3 c4 c5 c6 c7 c8 c9 cs cs1 cs2 cs3 cs4 cs5 cs6 cs7 cs8 cs9 s1 s2 s3
ŋʷ p
b
t
d
k
g
kʷ gʷ ƥ
ƃ
ʧ
ʤ f
v
s
z
h
hʷ r
j
w l
Figure 7: The Omuanwa subject performance in consonants.
3 Discussion
A close look at the performance of the subjects demonstrates that the majority of the
subjects recorded 100% accuracy in the production of 9 to 10 consonants, except the
Odeegnu subject that recorded only 2 consonants as shown in Table 17 and Figure 3 to
Figure 7.
Comparing the total performance of target sounds by the subjects, therefore, it is observed that the Emowha subject has acquired a greater percentage of 75.6 of the adult
sounds, followed by the percentage scores of 64.4 and 64.3 by the Aluu and Akpo subjects, respectively. The Omuanwa subject scored 56.5% and Odeegnu scored below average of 43.2%. While the reason for the low performance of the Odeegnu subject could
be attributed to age factor (3 years) that of the Omuanwa subject may be due to delayed
acquisition of the target or slight speech disorders. This, of course, requires further investigation before conclusion could be drawn.
153
Roseline I. C. Alerechi
Having noted the performance of individual subject in the target consonant, one may
say that the easy sounds for Ikwere subjects generally are [p t k g d b ʧ ʤ f j] particularly
if they occur in the target speech. On the other hand, [ƥ r h hʷ] appear more problematic
than other consonants. Some others not listed as either easy or difficult may be easy or
difficult based on the unique articulatory performance of the subject. Thus, the plosives,
affricates, fricatives and approximants seem not to constitute areas of difficulty, whereas
the implosives, tap and glottal fricatives do.
4 Conclusions
Inthis paper, we have been able to identify the various changes children impose on the
Ikwere (target) language as they articulate certain consonants. Some of these changes
conform to the forms used by speakers of different geographical areas, while others are
characteristic of child language. Adult speakers of Ikwere should be aware of the existing varieties of the language and the forms peculiar to children as this could facilitate
communication between children and adults, thereby, preventing problems engendered
by communication gap.
References
Akmajian, Adrian, Richard A. Demers, Ann K. Farmer & Robert M. Harnish. 2008. Linguistics: An introduction to language and communication. 5th edn. New Delhi: PrenticeHall of India.
Akpan, Ekaete E. 2004. Topics in developmental psycholinguistics. Calabar: Paico Press &
Books.
Akpan, Ekaete E. 2008. Vowel processes in the ibibio 2½ to 4½ years old monolingual children. In Shirley Yul-Ifode & Rotimi Badejo (eds.), Reading on child language and communication disorders in Nigeria, 239–250. Choba: University of Port Harcourt Press.
Akpan, Ekaete E. 2010. A study of the segmental phonology of Ibibio children. Journal
of Child Development and Communication Disorders 2. 151–170.
Alerechi, Roseline I. C. 2007a. A dialect survey of Ikwere: A phonological perspective.
Choba: University of Port Harcourt Doctoral dissertation.
Alerechi, Roseline I. C. 2007b. Labial variation in Ikwere. In Ozo-Mekuri Ndimele (ed.),
Convergence: English and nigerian languages, 347–369. Port Harcourt: Mand J Grand
Orbit Communication Ltd. & Emaih Press.
Alerechi, Roseline I. C. & G. Awala. 2012. Consonant substitution in the speech of the
Ekpeye child. Journal of Child Development and Communication Disorders 4. 249–260.
Alerechi, Roseline I. C. & Obed Ojukwu. 2010. Vowel processes in the speech of Ikwere
child. Journal of Child Development and Communication Disorders 2. 44–53.
Anthony, A., A. Bogle, Thomas T. S. Ingram & M. W. McIsaac. 1971. Edinburgh articulation
test. London: Longman Group.
Bolinger, Dwight. 1975. Aspects of language. New York: Harcourt Brace Jovanich.
154
9 Consonant substitution in child language (Ikwere)
Crystal, David. 1997. The Cambridge encyclopedia of language. Cambridge: Cambridge
University Press.
David, Stephen O. 2009. A cognitive perspective on language development of the Nigerian child. Journal of Child Development and Communication Disorders 10. 104–117.
Donwa-Ifode, Shirley O. & S. A. Ekwulo. 1987. Ikwere orthography. In Rebecca N. Agheyisi (ed.), Orthographies of Nigerian languages: Manual v. Lagos: National Language
Centre, Federal Ministry of Education.
Donwa-Ifode, Shirley O. & Nicholas Faraclas. 2001. A dialect atlas of Ikwere. Coba: Port
Harcourt University Press.
Fromkin, Victoria, Robert Rodman & Nina Hyams. 2003a. An introduction to language.
Australia: Thompson Wardsworth.
Fromkin, Victoria, Robert Rodman & Nina Hyams. 2003b. An introduction to language.
Australia: Thompson Wardsworth.
Hyman, Larry M. 1975. Phonology: Theory and analysis. New York: Holt, Reinhart & Winston.
Labarba, Richard C. 1981. Foundations of developmental psychology. New York: Academic
Press.
Menn, Lise. 1992. Acquisition of language: Phonology. In William Bright (ed.), International encyclopedia of linguistics 1. 12–16. New York: Oxford University Press.
O’Grady, William, John Archibald & Francis Katamba. 2011. Contemporary linguistics: An
introduction. Harlow: Pearson Education Limited.
Ojukwu, Obed & Roseline I. C. Alerechi. 2011. A process account of child phonological
system: Insights from Ohuhu-igbo. Journal of Child Development and Communication
Disorders 3. 65–76.
Williamson, Kay. 1980. Reading and writing Ikwere. Port Harcourt: Rivers Readers
Project.
Williamson, Kay. 1988. Linguistic evidence for the prehistory of the Niger-Delta. In Ebiegberi J. Alagoa, Frederick N. Anozie & Nzewunwa Nzenwa (eds.), The early history of
the Niger Delta, 65–119. Hamburg: Burske.
Williamson, Kay & Rogger Blench. 2000. Niger-congo. In Bernd Heine & Derek Nurse
(eds.), African languages: An introduction, 11–42. Cambridge: Cambridge University
Press.
Yul-Ifode, Shirley. 1999. A course in phonology. Choba: University of Port Harcourt Press.
Yul-Ifode, Shirley. 2003. Aspects of the phonology of six normally developing Ibibio children.
Poster presented at the symposium on research in child language disorders, University
of Wisconsin, Madison, USA, June 5th to 7th 2003.
Yul-Ifode, Shirley. 2008. The relation between language acquisition and language change
(evidence from Southern nigeria). In Shirley Yul-Ifode & Rotimi Badejo (eds.), Reading
on child language and communication disorders in Nigeria, 251–265. Choba: University
of Port Harcourt Press.
155
Chapter 10
A morphosyntactic analysis of adjectives
in two Kwa languages: Ga and Dangme
Regina Oforiwah Caesar
University of Education, Winneba, Ghana
Yvonne A. A. Ollennu
University of Education, Winneba, Ghana
The adjective category normally serves as attribute for the nouns in languages that do have
them. The paper investigates the morphosyntactic properties of adjectives in two Kwa languages, Ga and Dangme. Both languages have derived and non-derived adjectives. The paper which is mainly descriptive, examines the similarities and differences that exist between
these two Kwa languages in terms of their morphological and syntactic features. The paper
reveals that though similarities exist in the occurrence of adjectives syntactically, there exist differences in their morphological properties. On the other hand, Ga and Dangme show
agreement in terms of number with the head noun for all adjectives used attributively. The
paper concludes that in both languages, adjective occur after the head noun in attributive
position. Predication of adjectives can occur in nominal forms and the verbal equivalence
is also employed in both languages. Plural marking in adjectives is through reduplication
and affixation in Ga while in Dangme, it is only through affixation. Data for this paper were
collected from both primary and secondary sources.
1 Introduction
Adjectives as one of the subclasses of modifiers have been studied in Ghanaian languages
by several linguists (e.g., Osam 2003; Ameka 2003; Otoo 2005; Adjei 2007; Amfo et al.
2007; Danti 2007; Dzameshie 2007; Naden 2007; Pokuaa et al. 2007; Caesar 2013, a.o.). This
study investigates the morpho-syntactic properties of adjectives in Ga and Dangme. Ga
and Dangme belong to the Kwa group of languages from the Niger Congo family. Ga is
a two tone language whiles Dangme is a three tone language. Ga has twenty six letters
in its alphabet whiles Dangme has thirty letters of the alphabet. Both languages have
seven oral vowels and five nasal vowels.
Regina Oforiwah Caesar & Yvonne A. A. Ollennu. A morphosyntactic analysis of adjectives in two Kwa languages: Ga and Dangme. In Jason Kandybowicz, Travis Major, Harold
Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected papers from
the 45th Annual Conference on African Linguistics, 157–173. Berlin: Language Science Press.
DOI:10.5281/zenodo.1251726
Regina Oforiwah Caesar & Yvonne A. A. Ollennu
Ga is spoken along the coastal area in Accra. These areas include Ga Mashi, Osu, La,
Teshi, Nungua, Tema, Oyibi, Bawaleshi, and its surrounding villages. The area stretches
to the foot of the Akwapim hills, the Nyanam hill up to Ninobi, and then to the southwards of Langma hill in the southern part of Kasoa, whiles Dangme is spoken in two
regions of Ghana: Eastern and Greater Accra mainly in South-Eastern Ghana along the
coastal belt and the forest areas. Dangme speakers thus, inhabit the coastal area of the
Greater Accra Region, east of Accra and part of Eastern Region of Ghana. Dangme has
seven dialects while Ga has no dialects but there may be vocabulary differences geographically in terms of pronunciation. The seven dialects of Dangme include: Ada, Nugo,
Kpone, Gbugblaa/Prampram, Osudoku, Sɛ and Krobo (Yilo and Manya). There are several small communities east of the Volta Region that trace their origins to Dangme land;
most of these have shifted to Ewe as a language of daily life but others have not (Dakubu
1966, Sprigge 1969 cited in Ameka & Kropp Dakubu 2008). Patches of speakers are also
found in Togo land for instance, Nyetoe and Gatsi. Data used for this study were collected from some native speakers of Ga and Dangme and were cross-checked with other
native speakers. Data were also drawn from books (Ablorh-Odjidjah 1961; Kropp Dakubu
1987; 2000; Adams 1999; 2000; Amfo et al. 2007; Adi 2003; Odonkor 2004; Otoo 2005; and
Caesar 2013). The aim of this paper is to find out how similar or different the usage of
adjectives in these two languages looks like. It is said that a concept in a language may
be expressed in another, using a word from a particular word class but the same concept
may be expressed in another language using a word from a different word class (Dixon
1977; 1982; 2004). The theoretical framework is based on Dixon’s (2004) classification of
adjectives.
1.1 Theoretical framework
Dixon (2004) identifies a set of semantic types of property concepts that are encoded by
the adjective class in languages that have them. There are thirteen classes in his recent
work, which are:
• dimension e.g. big, small, long, deep, etc
• physical property, e.g. hard, strong, sweet, cheap, etc
• speed, e.g. fast, quick, rapid, etc
• age, e.g. new, old young, modern, etc
• colour, e.g. black, white, golden, etc
• value, e.g. good, bad, lovely, pretty, etc
• difficulty, e.g. easy, tough, hard, simple, etc
• volition, e.g accidental, purposeful, deliberate, etc
• qalification, (this has subtypes) e.g. true, obvious, normal, right, etc
• human propensity, (this also has subtypes), e.g. angry, jealous, clever, sad, etc
• similarity, e.g. different, equal (to) analogous (to), etc
158
10 A morphosyntactic analysis of adjectives in two Kwa languages: Ga and Dangme
• qantification, e. g. many, few, plenty, little, etc
• position, e.g. high, low, etc.
For the purpose of this study, we use examples from the following classes: dimension,
colour, value, age and physical property. This is because the adjectives in these two Kwa
languages are mostly found in these groups. Other semantic groupings will be investigated in future.
2 Sources of adjectives
Linguistic scholars (Kropp Dakubu 1987; Adi 2003; Adams 1999; 2000; Odonkor 2004;
Amfo et al. 2007; Caesar 2013) have identified that Ga and Dangme have both deep level
and derived adjectives. Deep level adjectives are monomorphemic, that is, they cannot
be segmented into morphemes to be meaningful. Examples of deep level are found below
in (1).
(1)
Ga:
agbo ‘big’
kpitioo ‘short’
kpakpa ‘good’
kakadaŋŋ ‘long’
gojoo ‘huge’
Dangme:
yumu ‘black’
kpiti ‘short’
kpakpa ‘good’
wayoo ‘small’
gojoo ‘huge’
2.1 Adjectives derived from verbs
Adjectives in Dangme could also be derived from verbs through either total or partial
reduplication. When adjectives are derived through total reduplication in Dangme, the
reduplicant takes an additional segment base on the vowel of the verbs stem, that is, verb
stems that end in {u, o, or ɔ} take {i, e or ɛ} to arrive at the adjective. Two processes occur
in partial reduplication: the deletion of a consonant and a rise in tongue high level of the
vowel of the reduplicant morpheme. Consider the following examples in Dangme:
(2) Dangme:
Verb
bla ‘to join’
ngla ‘to burn’
sa ‘to spoil’
tsu ‘to redden’
gbo ‘to die’
pɔ ‘to wet’
fi ‘to tie’
si ‘to fry’
ngma ‘to write’
Reduplicated form
ba∼blɛ ‘joint’
nga∼nglɛ ‘burnt
sa∼sɛ ‘rotten’
tsu∼tsu ‘red’
gbo∼gboe ‘dead’
pɔ∼pɔe ‘wet’
fi∼fii ‘tried’
si∼sii ‘fried’
ngma∼ngmɛɛ ‘written’
159
Regina Oforiwah Caesar & Yvonne A. A. Ollennu
The lateral /l/ is elided in the base of the reduplicated forms bablɛ ‘joint’ and nganglɛ
‘burnt’ and there is a rise in tongue height. That is, the low front vowel /a/, has changed
to the low mid vowel /ɛ/ as exemplified in the first two examples of (2) above. From
the data, it is observed also that reduplicated adjectives in Dangme can be formed by
copying the whole of the base and by adding a vowel to the reduplicated part of the verb
as shown in the last five examples of (2) above.
Adjectives could also be derived from verbs in Ga through affixation and reduplication.
Affixation is characterised with the suffixation of {-ŋ}, {-ru} and {-ra} to the base form of
some verbs to form adjectives. {-ŋ} is attached to verb stems that end in /i, ɛ/. {-ru} is
suffixed to verb stems that end in /u/, and {-ra} is attached to root forms that end in /a/.
Consider the Ga examples below:
(3)
Ga:
Verb
gbi ‘to dry’
di ‘to blacken’
yɛ ‘to whiten’
tsu ‘to ripe’
sha ‘to spoil’
Adjective
gbi-ŋ ‘dried’
di-ŋ ‘black’
yɛ-ŋ ‘white’
tsu-ru ‘red’
sha-ra ‘spoilt’
The Ga examples in (3) above have adjectives formed from verbs through affixation.
The illustrations below also show the reduplication process. The verb is either suffixed
with a segment or morpheme before reduplicated, or it is reduplicated and then suffixed
with the /i/ segment as in (4) below:
(4)
Ga:
boda ‘to be bent’
kwɔ ‘to be deep’
nyaŋe ‘to shun/despise’
boda∼bodai ‘crooked’
kwɔŋ∼kwɔŋ ‘deep’
nyaŋemɔ∼nyaŋemɔ ‘disgusting’
In complete reduplication in Ga, the segment /i/ is attached to the reduplicated part
of some verbs as in boda-boda-i ‘crooked’. However, a verb such as kwɔ ‘to be deep’ is
suffixed with the velar nasal /ŋ/ to become kwɔŋ before it is reduplicated as kwɔŋkwɔŋ
‘deep’. It is realised that the velar nasal, /ŋ/ is attached to the base as well as the reduplicated part of the adjective. Similarly, the verb root, nyaŋe ‘to shun/despise’ has been
suffixed with {-mɔ} to become nyaŋemɔ before it is reduplicated as nyaŋemɔnyaŋemɔ
‘disgusting’. This class of verbs is limited in Ga.
2.2 Adjectives derived from nouns
The stock of Dangme and Ga adjectives could also be added to through reduplication of
nouns. In Ga, the nouns are mostly pluralised first and then reduplicated. In Dangme,
however, some of the nouns just go through a total reduplication while others are pluralised as in Ga before they are reduplicated to derive adjectives. Consider the Dangme
examples below:
160
10 A morphosyntactic analysis of adjectives in two Kwa languages: Ga and Dangme
(5) Dangme:
Noun
zu ‘sand’
mamu ‘powder’
nyu ‘water’
zↄ ‘oil’
tso ‘tree’
tɛ ‘stone’
wu ‘bone’
kpↄ ‘lump’
Reduplicated form (Adjective)
zu∼zu ‘sandy’
mamu∼mamu ‘powdery’
nyu∼nyu ‘watery’
zↄ∼zↄ ‘oily’
tsohi∼tsohi ‘spongy’
tɛhi∼tɛhi ‘rocky’
wuhi∼wuhi ‘bony’
kpↄhi∼kpↄhi ‘lumpy’
It is observed in the Dangme examples in (5) that the first four examples of the reduplicated nouns have not taken on any affix after the reduplication. However, the last
three have attached the plural marker for common nouns, {-hi}, to indicate that they are
countable.
(6) Ga:
Noun
tɛ ‘stone’
tso ‘tree
nu ‘water’
shia ‘sand’
kpaa ‘rope’
kpɔ ‘lump’
wu ‘bone’
Reduplicated form (Adjective)
tɛi∼tɛi ‘stony/rocky’
tsei∼tsei ‘spongy’
nui∼nui ‘watery’
shia∼shiai ‘sandy’
kpaa∼kpai ‘knotty’
kpɔi∼kpɔi ‘lumpy’
wui∼wui ‘bony’
Ga generally attaches the plural suffix {–i} to count nouns while Dangme adds {-hi} to
common count nouns to form its plural. It is observed from the data on Ga that unlike
Dangme which pluralises only the count nouns, Ga pluralises the count and some noncount nouns by attaching the morpheme {-i} to the base and the reduplicated forms of the
word in many cases like nu ‘water’ becoming nu-i-nu-i ‘watery’. It is however observed
that the base form of the noun, shia ‘sand’ has not been pluralised in the reduplicated
form shia-shia-i ‘sandy’. A few of the nouns however, do not add the {-i} segment to the
nouns to form adjectives in Ga. See some examples in Ga below:
(7)
ŋmↄtↄ ‘mud’
kotsa ‘sponge’
ŋoo ‘salt’
ŋmↄtↄ∼ŋmↄtↄ ‘muddy’
kotsa∼kotsa ‘spongy’
ŋoo∼ŋoo ‘salty’
3 Morphosyntactic properties
This section discusses the morphological process of number agreement and reduplication
of adjectives in these two Kwa Languages. Adjectives in Ga and Dangme can inflect
for number (Kropp Dakubu 1987; 2000; Adams 1999; 2000). The Ga adjective is marked
161
Regina Oforiwah Caesar & Yvonne A. A. Ollennu
overtly to indicate plurality to show agreement with the head noun it modifies. However,
in Dangme, adjectives are not marked morphologically to show number agreement in
the constructions when the definite article is present in the NP. The number agreement
is marked on the definite article to indicate plurality and therefore has scope over the
entire noun phrase.
3.1 Plural formation in adjectives in Ga and Dangme
In Ga and Dangme, the plural affix of a noun, is either attached to the adjective that
qualifies the preceding noun or follows the NP. The adjective suffixes in Ga include {-i.
-ji. -bii} and the zero morpheme in few instances. Dangme also has noun plural suffixes
{-hi, -mɛ, -bi, -wi, -li}. The animate or human nouns in Dangme are what are marked
morphologically for plural in Dangme. In noun phrases where adjectives are present,
the adjectives select the {-hi} plural marker. {-mɛ} however, is attached to definite and
indefinite articles, such as {ɔ, a or ko}, to form the plural of the articles. The plural marking
on any of the articles scopes over the entire noun phrase. Consider the examples in
Dangme below:
(8)
Dangme:
Bo tsutsu-hi ngɛ daka a mi.
cloth red-pl are box def in
‘There are red cloths in the box.’
(9)
Mangoo mumu ɔ-mɛ sa.
mango fresh def-pl rotten.
‘The fresh mangoes are rotten.’
(10) Duku futa a-mɛ sɛ mu.
scarf white def-pl dirty.
‘The white scarves are dirty.’
In example (8), the adjective tsutsu-hi ‘red.pl’ has taken the plural marker of the noun
bo ‘cloth’ which is {-hi} retaining the noun in its singular form in the clause. In example
(9), the adjective mumu ‘fresh’ comes in between the subject noun mangoo ‘mango’ and
the definite plural marker ‘ɔmɛ’ in the morphology. Similarly in example (10), the adjective futaa ‘white’ comes in between the subject noun duku ‘scarf’ and the definite plural
marker, amɛ.
In Dangme we observe that to show number agreement in a noun phrase, it is the
plural form of the definite or indefinite articles that is used. The definite or indefinite
article informs us that the noun and adjective are in their plural forms even though the
nouns are not marked morphologically as in (8-10) above. Consider the Ga examples in
(11-13).
162
10 A morphosyntactic analysis of adjectives in two Kwa languages: Ga and Dangme
(11) Ga:
Wo-ji he-i
lɛ dara.
book-pl new-pl def big.iter
‘The new books are big.’
(12) Atade-i fɛɛfɛ-ji
lɛ elaaje.
dress-pl beautiful-pl def lost
‘The beautiful dresses are lost.’
(13) Tse-i kakada-ji lɛ kumɔ.
tree-pl long-pl def broke.iter
‘The tall trees broke.’
On the other hand, Ga has a suffix on both the nouns wo-ji ‘books’ atade-i ‘dresses’
and adjectives he-i ‘new’, fɛɛfɛ-ji ‘beautiful.pl’, tse-i ‘tress’, and kakada-ji ‘tall ones’ to
show number agreement. The definite article however, is not marked as it has no plural
form in Ga. Another observation from the above Ga example is that the definite article
retains its shape for both the singular and the plural forms of the nouns. Some adjectives
are also reduplicated in Ga and Dangme to express plural number in the entity named
as in (14-17) below:
(14)
Ga:
E-he
shikpɔŋ lɛkɛtɛɛ.
3sg-buy land
wide
‘He/she bought a wide land.’
(15)
E-he
shikpɔ-ji lɛkɛtɛ∼lɛkɛtɛɛ.
3sg-buy land-pl wide∼red
‘He/she bought wide lands.’
(16) Dangme: E jua
blodo daka.
3sg sell.aor bread box
‘He/she sold a box full of bread.’
(17) E juaa
blodo daka∼daka.
3sg sell.aor bread box∼red
‘He/she sold boxes of bread.’
The Ga example lɛkɛtɛɛ ‘wide’ in (14) denotes a singular number. On the other hand,
lɛkɛtɛ-lɛkɛtɛɛ ‘wide-wide’, denotes plurality in the plural noun, shikpɔ-ji it qualifies in
(15). But in Dangme, the mere reduplication of the adjective does denote plurality. Plural
marking affixes are however, not expressed on the noun being modified as exemplified
in (16-17).
163
Regina Oforiwah Caesar & Yvonne A. A. Ollennu
3.2 Reduplication of adjectives
Ga and Dangme adjectives can also be reduplicated. Normally when an adjective is reduplicated, it shows intensity. As in the NPs, the reduplicated adjectives are pluralised in
both their base and the reduplicant parts. Below are examples to illustrate:
(18)
Ga:
wulu ‘big’
kpitioo ‘short’
bibioo ‘small’
wamaa ‘large’
(19) Dangme:
agbo ‘big’
nyafi ‘small’
yumu ‘black’
tsutsu ‘red’
wuji∼wuji ‘big’
kpitibii∼kpitibii ‘short’
bibii∼bibii ‘small’
wamaa∼wamaa ‘large’
agbo∼agbo ‘big’
nyafi∼nyafi ‘small’
yumu∼yumu ‘blackened’
tsutsu∼utsu ‘reddish’
The reduplication process can be total or partial. This is demonstrated in example (19)
where tsutsu ’red’ becomes tsutsuutsu ’reddish’. The reduplication process in Dangme is
total whereas in Ga, there is generally the suffixation of {-i, -bii, -jii} as plural affixes to adjectives and nouns. With the exception of few adjectives such as wamaa –wamawamaa,
all other nouns and adjectives take any of the three suffixes above. Their conditioning is
not discussed in this paper. See below how some of these reduplicated forms of adjectives
can occur in sentences.
(20) Ga:
Gbekɛ bibioo lɛ e-wↄ.
child small def perf-sleep
‘The little child is asleep.’
(21)
Gbekɛbii bibii∼bibii lɛ e-wↄ.
child-pl small∼very def .perf-sleep
‘The little children are asleep.’
(22) Dangme:
Tade yumu ɔ gba.
dress black def tear.aor
‘The black dress is torn.’
(23)
164
Tade yumu∼yumu ɔ ngɛ tsu ↄ mi.
dress black∼very def is room def inside
‘The very black dress is in the room.’
10 A morphosyntactic analysis of adjectives in two Kwa languages: Ga and Dangme
In the Ga examples in (20-21), it is observed that the reduplicated form of bibioo ‘a
small…’ is bibiibii ‘very small ones’. Bibioo expresses singularity in the entity being discussed while bibiibii expresses intensity and plurality. In the Dangme examples in (2223), it is observed that yumu ‘black’ has been reduplicated as yumuyumu ‘very black’.
The reduplicated form, yumuyumu, shows the intensity of the colour yumu, ‘black’.
Both Ga and Dangme nominalise the adjective to be the head of an NP and also to be
a subject of a sentence. To nominalise an adjective in Ga and Dangme, the prefix {e-} is
attached to certain class of adjectives. However some of these adjectives are not attached
with the prefix but remain in the same form as nominals. The prefix {e-} nominalises a
class of adjectives that denotes colour and age of objects. For instance, in Ga, yɛŋ ‘white’
becomes e-yɛŋ ‘white one’, diŋ ‘black’ becomes e-diŋ ‘black one’, tsuru ‘red’ becomes etsuru ‘red one’, hee ‘new’ becomes e-hee ‘new/new one’, momo ‘old’ becomes e-momo ‘old
one’, ŋmↄŋ ‘fresh’ becomes e-ŋmↄŋ ‘fresh one’. Likewise, in Dangme, ku ‘male’ becomes
ku-e-ku ‘male one’, yo ‘female’ becomes e-yo ‘female’, he ‘new’ becomes e-he ‘new one’,
agbo ‘big’ becomes e-agbo ‘big one’, wayoo ‘small’ becomes e-wayoo ‘small one’. The
conditioning for the zero allomorph is yet to be investigated.
4 Functions of adjectives
Dixon (2004) asserts that adjectives typically fill two roles in the grammar of a language.
These two roles are the attributive and predicative use of adjectives. In addition to these
roles, the adjective can occur in comparative constructions. When an adjective plays the
attributive role, it serves as a modifier to the head noun. When the adjective is used
predicatively, it occurs as a copula complement in most languages. Dixon (2004) notes
however that these two roles, attributive and predicative may not occur for all adjectives in all languages. In certain instances, only one of these roles may be found. Such
adjectives can occur within sentences as exemplified below. We begin with adjectives in
attributive position.
4.1 Adjectives in attributive position
Dangme adjectives as well as Ga adjectives that qualify nouns have the reversal structure
as compared to English. Such adjectives come after the nouns they qualify in a phrase,
clause or of any kind. Consider the following examples in Dangme and Ga:
(24)
Dangme:
Womi he ↄ ka.
book new def be.long
‘The new book is long.’
(25)
Bɔɔlu momo ɔ pɛ.
ball old
def burst
‘The old ball is burst.’
165
Regina Oforiwah Caesar & Yvonne A. A. Ollennu
(26)
Kɔɔpoo agbo ɔ hyi.
cup
big def be.full
‘The big cup is full.’
(27)
E
juaa
blodo bɔdɔɔ.
he/she sell.hab bread soft
‘He/she sells soft bread.’
(28)
Ga:
Nuu agbo lɛ wɔ vii.
man big def sleep. pst deeply
‘The big man slept soundly/deeply.’
(29)
Wolo hee lɛ da.
book new def be big
‘The new book is big.’
(30)
Atade momo lɛ e-tsere.
dress old
def 3sg-tear
‘The old dresses are torn.’
(31)
Blodo bɔdɔɔ e-hɔ-ɔ.
bread soft 3sg-sell-hab
‘He/she sells soft bread.’
In examples (24-25) in Dangme, he ‘new’ and momo ‘old’ are expressing age. Agbo ‘big’
indicates dimension in (26) and bɔdɔɔ ‘soft’ shows physical property. All these adjectives
come after the nouns, womi ‘book’, bɔɔlu ‘ball’, kɔɔpoo ‘cup’ and blodo ‘bread’ in (27)
they respectively qualify. In example (28-31) on Ga, it is realised also that agbo ‘big’,
bɔdɔɔ ‘soft’, hee ‘new’ and momo ‘old’ express dimension, physical property, and age as
in the Dangme examples in (24-27). These adjectives in Ga, are preceded by the head
nouns nuu, ‘man’, wolo ‘book’, atade ‘dress’ and blodo ‘bread’. From the above examples
in (24-27) in Dangme and (28-31) in Ga, we see the adjectives occurring after the nouns
they modify, followed by the definite article, if present in the clause. Thus, the adjective
comes in between the noun and the definite or indefinite article as in other Ghanaian
languages.
The adjective for ‘beautiful’ in Ga is fɛɛfɛo. In example (32), fɛɛfɛo is used attributively
in a clause. Dangme on the other hand, employs a whole phrase such as ‘he ngɛ fɛu’
or ‘kɛ e he fɛu’ in other to express the attributive use of the adjective, ‘beautiful’ as in
example (33) below:
(32)
166
Ga:
Asupaatere fɛɛfɛo lɛ etse.
shoe/sandals beautiful def tear.pst
‘The beautiful pair of shoes is torn.’
10 A morphosyntactic analysis of adjectives in two Kwa languages: Ga and Dangme
(33)
Dangme:
Tokota
kɛ e
he fɛu
ɔ hia.
shoe/sandal with 3sg.poss part beauty def tear.aor
‘The beautiful pair of shoes is torn.’
4.2 More than one adjective in attributive position
The paper also investigates when more than one adjective is used attributively for a noun.
In Ga, it is observed that sometimes the last adjective used in a sequence is prefixed with
{e-}. Consider the Ga examples (34-37) below:
(34) Ga:
ko ka jɛmɛ.
Tɛ bibioo e-hee
stone little nmlz-new def lie there.
‘There is a new small stone lying there.’
(35)
Gbee agbo kpitioo lɛ gbo.
dog big short def die.pst
‘The big short dog is dead.’
(36)
Mi-he
shia fɛɛfɛo agbo ko.
1sg-buy.pst house beautiful big certain.
‘I bought a certain big beautiful house.’
lɛ egbo.
(37) Gbe-i agbo-i e-di-ji
dog-pl big-pl nmlz-black-pl def perf.die
‘The big black dogs are dead.’
In the Ga examples in (34-37), the adjectives are bibioo ‘small’ and ehee ‘new’, agbo
‘big’ and kpitioo ‘short’, fɛɛfɛo ‘beautiful’ and agbo ‘big’, agbo-i ‘big ones’ and edi-ji ‘black
ones’ have respectively occurred in a sequence. In Dangme, the category of age, value or
colour may precede those with physical or dimension properties. Those from the physical
property and human property tend to be used in copula complement function. Sentences
(38-41) present some examples of adjective sequencing in Dangme.
(38)
Dangme:
Mangoo ngmlikiti gaga nↄ si.
mangoo ripe.neg long fall down.
‘The unripe oval shaped mango has fallen down.’
(39) To
futa agbo kɛ e
nane gagaaga a laa.
sheep/goat white big with 3sg.poss leg long.red def lost
‘The white fat sheep/goat with the long legs is lost.’
167
Regina Oforiwah Caesar & Yvonne A. A. Ollennu
(40) Sɛ
yumu nyafii ngɛ sukuu tsu ↄ mi.
Stool/chair black dim is school room def inside
‘The black small stool/chair is in the classroom.’
(41)
Lo
momo sasɛ ↄmɛ ngɛ tso goga a mi.
fish/meat old
rotten def.pl at wood bucket def inside
‘The little rotten old meat/fish is in the wooden bucket.’
4.3 Predicative use of the adjective
The paper now examines whether the adjective can function predicatively in the two languages. Dixon (2004: 106) asserts that adjectives function as copula complement usually
referred to as predicative adjectives. A predicatively used adjective is one kind of subject
complement. It is an adjective that modifies the subject of the sentence. In addition, they
function as adjectives usually to qualify the NPs they occur with within sentences as
exemplified below:
(42) Ga:
Asapaatere lɛ yɛ
fɛo.
Shoe/sandals def possess beauty
‘The pair of shoe/sandals is beautiful.’
(43)
Dangme:
Tokota
a ngɛ
fɛu.
Shoe/sandals def possess beauty.
‘The pair of shoe/sandals is beautiful.’
In the examples (42) and (43) the use of the adjectives in complement position are normally nominalised. The copula verb yɛ and ngɛ are used respectively in Ga and Dangme.
The adjective, fɛo in (42) and fɛu in (43) are the nominal forms of the adjectives in Ga and
Dangme. The Ga adjective normally does not occur in predicate position except when
prefixed with {e-} or remains in the same form (zero morpheme). Sometimes adjectives
which have verbal equivalents are used by native speakers of both Ga anad Dangme
languages. The constructions below indicate this phenomenon:
(44) Dangme:
Siadeyo ka.
Siadeyo be.tall
‘Siadeyo is tall.’
(45)
168
Bo ɔ pɔ.
cloth def be.wet
‘The cloth is wet.’
10 A morphosyntactic analysis of adjectives in two Kwa languages: Ga and Dangme
(46)
Tso ɔ gbli.
tree def be.dry
‘The tree is dried.’
(47)
Ga:
Aku kwɔ.
Aku be.tall
‘Aku is tall.’
(48)
Atade lɛ e-fɔ.
dress def perf-wet
‘The dress is wet.’
(49) Tso lɛ gbi.
tree def perf.dry
‘The tree is dry.’
Examples (44-46) are intransitive clauses with one core argument each in the subject
position. The morphemes in bold print are intransitive predicates with their heads being
verbs. All verbs which denote adjectival meanings in sentence (44-46) occupy the predicative position but modify the noun Siadeyo ‘a personal name’, bo ‘cloth’ and tso ‘tree’
in the Dangme clauses. In a similar way, kwɔ ‘be tall’, fɔ ‘wet’ and gbi ‘dry’ modify the
nouns Aku, ‘a personal name’, atade ‘dress’ and tso ‘tree’ in the Ga examples in (47-49)
above.
From the examples in (44-49), it was realised that some of these adjectives when used
predicatively make use of verbs when there are equivalence in Ga and Dangme. It was
also noted that in Ga, when the verb equivalence is absent, there is the copula construction and the nominalised form of the adjective is used. When the adjectives are from the
human propensity class they normally tend to be nouns. Relative clauses are also sometimes used in both languages. Below are some other verbs that could be used to express
adjectival meanings.
(50) Dangme:
tí ‘ to be thick’
fú ‘to be ripe’
gbó ‘to die’
kle ‘to be big’
hì ‘to be good’
(51)
Ga:
ti ‘to be thick’
tsu ‘to be ripe’
gbo ‘to die’
da ‘to be big’
hi ‘to be good’
These are verbs as they take all aspect and tense markers like other verbs. Dixon (2004)
refers to such adjectives as verbal adjectives. These can be negated as shown in (52) and
(53). In Ga and Dangme, each of these takes a negative suffix. For instance:
169
Regina Oforiwah Caesar & Yvonne A. A. Ollennu
(52)
Dangme:
ti we ‘not thicken’
fu-i ‘not ripe’
gbo we ‘not dead’
klee we ‘not big’
(53) Ga:
tii-i ‘not thicken’
tsuu-u ‘not ripe’
gboo-o ‘not dead’
daa-a ‘not big’
In Ga, there are two verb classes, one class takes circumfixes most often to mark negativity and aspect and the other class employs suffixes. These have shown up in (53) where
the affirmative forms of the verbs are either suffixed or circumfixed with the negative
marker(s). That is, verbs of this category can also take on the full range of tense/aspect
markers just like any other verb in Ga and Dangme. According to Dixon (2004: 19), when
an adjective occurs as an intransitive predicate, it may take some or all the morphological processes available to verbs in the slot, thus, tense, aspects, mood, polarity, etc. An
adjective in Dangme functions directly as a modifier of a noun in a noun phrase, acting
as a copula complement and shows morphological categories similar to those of nouns
especially those related to numbers.
5 Similarities and differences
Ga and Dangme both have deep level adjectives which are monomorphemic. Both languages have derived adjectives from verbs and nouns which are obtained through total
or partial reduplication and an addition of a segment. In the process of reduplication,
Dangme attaches {-i, -e, or -ɛ} to verb stems that end in {-u, -o or -ↄ } while Ga attaches
{-ŋ, -ru, -ra} to verb stems that end in {-i, -ɛ, -u, -a} to derive adjectives. We found that in
the process of deriving adjectives from nouns in Ga and Dangme involves two processes.
That is, either the root noun goes through a complete reduplication only, or reduplicates
completely and then attaches a plural marker to both the root noun and the reduplicated
parts of the noun. Dangme attaches the {-hi} plural marker while in Ga, {-i} is attached
as discussed in examples (5-7). It is to be noted that whilst in Dangme, non-countable
nouns are not pluralised either in a part or in all the morphemes of the reduplicated
form of the noun, Ga does attach a plural affix, {-i} to each of the free forms of countable
and non-countable nouns reduplicated to form an adjective, with the exception of a few
that do not pluralise the root noun of the reduplicated word as in shia-shia-i ‘sandy’.
Adjectives can also be reduplicated to show intensity and express plural number in both
languages.
Adjectives in Ga and Dangme function attributively. It is to be noted that just as it
occurs in other Ghanaian languages such as Akan, Ewe, Gurene, Dagbani and the like,
Ga and Dangme adjectives and nouns have a reversed structure in the NP as compared
to English, i.e. the adjective occurs after the noun it qualifies. It is also to be noted that
while Ga has lexical nominal forms of adjectives such as fɛɛfɛo for the English word
‘beautiful’, Dangme on the other hand employs phrases such as ‘…he ngɛ fɛu’ or ‘kɛ e he
fɛu’.
170
10 A morphosyntactic analysis of adjectives in two Kwa languages: Ga and Dangme
It was realised that plural adjectives are marked morphologically on the adjective and
not on the noun stem in Dangme when the definite article is absent in a phrase, a clause
or a sentence. It is also noted that, where determiners come into play, the adjectives
come in between the noun and the determiner. However, the definite article in Ga does
not take any plural affix whiles the definite article in Dangme have plural form. The
definite article denotes number concord in the noun and adjective in Ga and Dangme.
When there is more than one adjective in a construction in Ga, they are all marked to
express plurality. In Dangme however, it is only the final adjective that takes the plural
suffix.
The study has identified that predicative adjectives are preceded by copula verbs in
both languages. While Ga employs the copula verb, yɛ, Dangme uses ngɛ. In Ga, when
the adjective is used predicatively, it is prefixed with {e-} or may occur in the same form.
Sometimes, verbs which denote adjectival meanings are used. Where there is no verbal
equivalence, the construction becomes a copula one.
6 Conclusion
This paper has examined certain morphological and syntactic features of adjectives in
Ga and Dangme, two Kwa languages spoken in Ghana. It considered the similarities and
differences between adjectives in Ga and Dangme, It was observed that the membership
for the adjective class in these two languages is increased by deriving adjectives from
other sources such as reduplication and the derivation of nouns and verbs. Although,
Ga and Dangme employ these three ways of forming adjectives, some of the processes
vary. When adjectives are derived through total reduplication, the reduplication template takes an additional segment base on the vowel of the verbs stem to express adjectival properties in Ga and Dangme. It was identified that the two languages have the
potential of using reduplicated verbs and nouns as adjectives to modify other nouns in
the language. Adjectives can also be reduplicated to show intensity as in other languages
of the world. In dealing with categories of adjectives, one can see that adjectives in Ga
and Dangme help differentiate one nominal from the other as in other languages.
The study established that even though Ga and Dangme are related to a large extent
in the area of adjectives, there are identifiable differences in their morphological and
syntactic properties of adjectives as discussed under the similarities and differences in
§5. It is hoped that this study will add to the typological study on adjectives universally.
Abbreviations
1
3
aor
def
hab
nmlz
first person
third person
aorist
definite article
habitual
nominalizer
np
perf
pl
poss
pst
sg
noun phrase
perfective
plural
possessive
past tense
singular
171
Regina Oforiwah Caesar & Yvonne A. A. Ollennu
Acknowledgements
We are grateful to Management of the University of Education, Winneba for permitting
us to attend and present this paper at ACAL45 in Kansas, USA.
References
Ablorh-Odjidjah, J. R. 1961. Ga wiemↄ ekomɛi a nitsumↄi. Accra: Presby.
Adams, George. 1999. Aspectuals in Ga. Accra: University of Ghana. Unpublished thesis.
Adams, George. 2000. Nominalization processes in Ga. Accra: University of Ghana. Unpublished thesis.
Adi, Daniel B. 2003. Animosa sua. Winneba: Teye-Ngua Publications.
Adjei, Francisca A. 2007. Adjectives in Siyase (Avatime). In George Dakubu Mary Esther Kroppand Akanlig-Pare, Emmanuel K. Osam & Kofi K. Saah (eds.), Proceedings of
the annual colloquium of the Legon-Trondheim linguistics project, 2005 (Studies in the
languages of the Volta basin 4), 122–134. Accra: University of Ghana Department of
Linguistics.
Ameka, Felix. 2003. The adjective class in Ewe: Its strata and emergent nature. Unpublished
manuscript.
Ameka, Felix K. & Mary Esther Kropp Dakubu. 2008. Aspect and modality in Kwa languages. Amsterdam: John Benjamins Publishing Company.
Amfo, Nana Aba, Stella Boateng & Yvonne A. Otoo. 2007. A comparative study of the
morphosyntactic properties of adjectives in three Kwa languages. In George Dakubu
Mary Esther Kroppand Akanlig-Pare, Emmanuel K. Osam & Kofi K. Saah (eds.), Proceedings of the annual colloquium of the Legon-Trondheim linguistics project, 2005 (Studies in the languages of the Volta basin 4), 60–71. Accra: University of Ghana Department of Linguistics.
Caesar, Regina O. 2013. Adjectives in Dangme. In Felix A. Fabunmi & Akeem S. Salawu
(eds.), Readings in African dialectology and applied linguistics, 39–50. München: Lincom.
Danti, Alexis. 2007. The Kasem adjective phrase structure. In George Dakubu Mary Esther Kroppand Akanlig-Pare, Emmanuel K. Osam & Kofi K. Saah (eds.), Proceedings of
the annual colloquium of the Legon-Trondheim linguistics project, 2005 (Studies in the
languages of the Volta basin 4), 116–126.
Dixon, R. M. W. 1977. Where have all the adjectives gone? Studies in Language 1. 19–80.
Dixon, R. M. W. 1982. Where have all the adjectives gone and other essays in semantics.
Berlin: Mouton.
Dixon, R. M. W. 2004. Adjective classes in typological perspective. In R. M. W. Dixon
& Alexandra Y. Aikenvald (eds.), Adjective classes: A cross linguistic typology, 1–49.
Oxford: Oxford University Press.
Dzameshie, Alex. 2007. A comparative study of adjectives in Ewe and Akan. In George
Dakubu Mary Esther Kroppand Akanlig-Pare, Emmanuel K. Osam & Kofi K. Saah
172
10 A morphosyntactic analysis of adjectives in two Kwa languages: Ga and Dangme
(eds.), Proceedings of the annual colloquium of the Legon-Trondheim linguistics project,
2005 (Studies in the languages of the Volta basin 4), 72–84.
Kropp Dakubu, Mary Esther. 1987. The Dangme language. Accra: Unimax Publishers Limited.
Kropp Dakubu, Mary Esther. 2000. The Ga nominal phrase. Unpublished manuscript.
Naden, Tony. 2007. Descriptives: Adjectives in Mampruli. In George Dakubu Mary Esther Kroppand Akanlig-Pare, Emmanuel K. Osam & Kofi K. Saah (eds.), Proceedings
of the annual colloquium of the Legon-Trondheim linguistics project, 2005 (Studies in
the languages of the Volta basin 4), 85–99. Accra: University of Ghana Department of
Linguistics.
Odonkor, Padi. 2004. An introduction to the grammar of Dangme: A linguistics approach.
Accra: Adjwensa Publishers.
Osam, Emmanuel Kweku. 2003. An introduction to the verbal and multiverbal system of
Akan. In Martha Larsen, Dorothee Beermann & Lars Hellan (eds.), Proceedings of the
Workshop on Multi-Verb Constructions. Trondheim: Norwegian University of Science
& Technology.
Otoo, Yvonne A. 2005. Aspects of Ga syntax: Adjectives. Trondheim: University of Trondheim dissertation.
Pokuaa, Christiana, E. Kweku Osam & Kofi K. Saah. 2007. Adjective sequencing in Akan:
A preliminary investigation. In George Dakubu Mary Esther Kroppand Akanlig-Pare,
Emmanuel K. Osam & Kofi K. Saah (eds.), Proceedings of the annual colloquium of the
Legon-Trondheim linguistics project, 2005 (Studies in the languages of the Volta basin
4), 140–149.
173
Chapter 11
Towards a unified theory of
morphological productivity in the Bantu
languages: A corpus analysis of
nominalization patterns in Swahili
Nick Kloehn
University of Arizona
Models arguing for a connection between morphological productivity and relative morpheme frequency have focused on languages with relatively low average morpheme to word
ratios. Typologically synthetic languages like Swahili which have relatively high average
morpheme to word ratios present a challenge for such models. This study investigates the
process of agentive nominalization from the perspective of the Dual Route Model. The findings suggest that all agentive nominal forms should decompose when accessed and thus that
speakers of Swahili should include these morphemes in their lexical inventory apart from
root morphemes. This process appears to not be influence by noun classification, or verbal
derivation.
1 Introduction
Within the realm of morphological processing in lexical access, a growing body of
research has been conducted with the aim of developing a quantitative theory of morphological productivity. The central notion of this body of research is the proposition
that a morpheme’s tendency to productively affix to a novel word-form is determined by
whether existing words containing said affix undergo morphological decomposition
when accessed in the lexicon (Bybee 1995, Hay 2002, Hay & Baayen 2002). However, studies supporting these findings have been conducted in Indo-European languages which
contain relatively low average morpheme per word ratios compared to the Bantu Languages.1 This study aims to kick start a vein of research investigating the quantitative
1
2.55 morphemes per word for Swahili versus 1.68 morphemes per word for Modern English (Greenberg
1959).
Nick Kloehn. Towards a unified theory of morphological productivity in the Bantu languages:
A corpus analysis of nominalization patterns in Swahili. In Jason Kandybowicz, Travis Major,
Harold Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected papers
from the 45th Annual Conference on African Linguistics, 175–190. Berlin: Language Science Press.
DOI:10.5281/zenodo.1251728
Nick Kloehn
determinants of morpheme productivity in a group of languages which have higher average morpheme to word ratios than those previously studied. This first step is made by
analyzing the process of nominalization in Swahili from the perspective of the dual
route model (Baayen 1992) (Henceforth, DRM) in order to determine whether derived
nominal morphology is predicted to be productive. The derivational process of nominalization is chosen by virtue of being analogous to the sorts of derivational patterns in
English used as evidence to support DRM. However unlike English, nominalization in
Swahili interacts both with noun classification and verbal derivation. This study touches
upon the interaction of these morphological systems to illustrate the complexity of the
issue of productivity in synthetic language. These systems are reviewed in §1.1 and §1.2.
The study is contextualized with a discussion of the relevant Lexical Access literature in
§1.3 as well as models of morphological productivity in §1.4 The history of morphological
processing in synthetic languages, and the motivation for the current study is outline in
§1.5. In §2 the paper outlines two corpus study designed to investigate the degree of predicted productivity of derived nominals in Swahili and the degree to which they interact
with noun classification. In §3 the degree to whether or not productivity interacts with
presence and number verbal extensions is evaluated. Finally, the results of these studies
are discussed and future research is suggested in §4.
1.1 Nominal and verbal derivation in Swahili
One of the primary features common to all members of the Bantu language subgroup
of the Niger-Congo language family is the presence of a rich noun class system (Heine
1982). In Swahili, noun class membership is indicated by a word initial prefix with little
exception. These prefixes determine the grammatical number and semantic distinction
of the nominal forms to which they are affixed. All nominals in Swahili must occur
in some noun class even when the noun class is not signaled by a prefix. Regardless
of whether the prefix is present, class membership can be determined by agreement
patterns of agreeing prepositions, verbal inflection, nominal phrase constituents, and
relativizers (Mohamed 2001). Example (1) gives the morphological breakdown for three
nominals in Swahili each corresponding to a different noun class. The degree to which
noun classification can be considered derivational or inflectional is debatable, but from
the purposes of the present study it is only important to note that the process of noun
classification has both inflectional properties (bound and obligatory) and derivational
properties (changes meaning and category).
(1)
(Swahili noun classes (Mohamed 2001, TUKI 2001))
a. mi-parachichi
cl4-avocados
‘avocado trees’
b. vi-atu
cl11-shoe
‘shoes’
176
11 Towards a unified theory of morphological productivity in the Bantu languages
c. u-nywele
cl8-hair
‘hair’
Another morphological process that is present in Swahili is the system of verbal extensions. These suffixes denote a type of derivation that alters the semantic denotation
of a verb and often the adicity of the verb undergoing this process.2 Henceforth, a verb
lacking a verbal extension will be described as simplex, and one which contains a verbal extension will be described as complex. Swahili takes advantage of its morphological system to represent semantically complex verbal environments in the form of morphologically complex verbs, whereas languages like English denote functionally analogous environments by both morphological and non-morphological means. For example,
verbs that are semantically analogous in English to those derived through the process
described in Example (2) must either (i) undergo the addition of derivational morphology
(e.g. lock - un-lock), (ii) undergo the addition of inflectional morphology (e.g. pay - paid),
(iii) introduce a phonologically and morphologically unrelated lexical item (e.g. close grasp), or (iv) require a bi-clausal structure (e.g. cook - make cook). Furthermore, verbal
extensions may occur multiple times forming multimorphemic complex verbs such as
Causative-Applicatives.3 Example (2) gives the derivation of two complex verbs (right
of the arrow) and their simplex counterparts (left of arrow).
(2)
(Complex verb derivations (Mohamed 2001, TUKI 2001, Seidl & Dimitriadis
2003))
a. yunj-a ↔ yunj-ik-a
break-fv break-stat-fv
‘break’ ‘be broken’
b. song-a ↔ song-o-a
press-fv press-aug-fv
‘press’ ‘press out’
Given that nominals in Swahili are mostly derived from verbs, the presence of verbal
extensions affects the amount of morphology contained within many nominal forms. In
other words, since nominals can be derived from both simplex and complex verb forms,
the issue of whether a verb is simplex or complex is relevant to any investigation of
nominalization. Regardless of whether a nominalized form is simplex or complex, the
nominal will obligatorily be classified in a given noun class.
1.2 Nominalization in Swahili
In addition to the noun class prefix and optional verbal extension, the derivation of deverbal nouns requires the addition of one of three word final suffixes denoting nomi2 I.e.
some extensions (e.g. Causative) require the addition of an argument of the verb and others (e.g. Reciprocal) require the subtraction of an argument of the verb.
3 E.g. anz-il-isha start-appl-caus ‘initiate’
177
Nick Kloehn
nalization (Katamba 2003). In Swahili, nominalizing suffixes denote the relationship of
the derived noun to the verb as either agentive or instrumental. Critically, agentive and
instrumental forms are limited in the noun class prefixes which it may take,4 as seen in
Table 1.5
In addition, Table 1 shows that deverbal nouns may be formed from both simplex and
complex verbs, and that they may also undergo both agentive and instrumental nominalization. Unfortunately, this paradigm does not cover the whole story. There also exist
deverbal nouns for which there is a morphological variant suffix -i. 6 However, the current research limits the domain of inquiry to the suffix denoting agentive nominalization
-aji which is loosely equivalent to the English suffix -er indicating the doer of an action.
1.3 Morphological processing in Lexical Access
The aim of the current subsection is to contextualize this study by reviewing the literature on how morphologically complex words are accessed. Research in the area of Lexical
Access is able to rely upon the fact that frequent words are accessed in the lexicon faster
than infrequent words. This word freqency effect (Broadbent 1967) has been used
to study morphological complexity as a methodology for determining whether affixed
words are stored together or separately in the lexicon. Taft (1979) found in a visual word
recognition task that when multi-morphemic words of a constant frequency vary in
the frequency of their stem morpheme, there was an asymmetry in time it took to access
these words. Specifically, a word like size-d, which contains a stem of high frequency
(size) was accessed faster than a word which has an identical surface frequency (rake-d),
but which contain a less frequent stem (rake). This suggests, Taft concludes, that words
containing a given stem are stored together in the lexicon (i.e. size is stored with sized,
sizable, sizing, etc.). Similarly, it has been suggested that stems that occur with more
non-inflectional affixes (e.g. calculate: calculable, calculation, calculator, calculus, incalculable, incalculably, miscalculate, and miscalculation) are accessed faster than those cooccurring with only few non-inflectional affixes (roar: uproar). This line of research has
shown that reaction times in Visual Lexical Decision tasks are faster for nouns having
a higher number of morphological neighbors for both mono-morphemic (Baayen et al.
2007, De Jong IV et al. 2000), and multi-morphemic nouns (Bertram et al. 2000)
Another vein of research has suggested that whether or not a complex word (e.g. happiness) is stored as one unit (happiness) or as multiple units (happy+ness) has to do with
whether or not the component morphemes are parsed in perception (Baayen 1992 and
Hay 2002). If one actively decomposes these forms during perception, then both units
will be stored in the lexicon (as happy and -ness), and if one does not actively decompose
them, then the form will be stored as a single unit (as happiness). Crucially, they claim
that parsing is a function of the frequency ratio of the individual units. Specifically, parsing occurs when the frequency of the stem (happy) is greater than the frequency of the
4
Class 3/4 may only occur with Instrumental Nominalization, and Class 1/2 may only occur with Agentive
Nominalization.
5
The associated verb is listed to the left.
6 E.g. mw-anz-il-ish-i cl1-start-appl-caus-agent_nom ‘founder’
178
11 Towards a unified theory of morphological productivity in the Bantu languages
Table 1: List of complex and simplex verbs and their corresponding deverbal nominal. These forms are divided by whether they are agentive or instrumentive nominals, and vary in their noun class. (Mohamed 2001,TUKI 2001,
Katamba 2003).
Deverbal Nouns
Simplex Verb
cheza
‘play’
pepe
‘fan’
Singular
Plural
Agentive Nominalization
of Class 1/2
m-chez-aji
cl1-play-agent_nom
‘player’
wa-chez-aji
cl2-play-agent_nom
‘players’
Instrumentive
Nominalization of Class
3/4
m-chez-o
cl3-play-inst_nom
‘game’
mi-chez-o
cl4-play-inst_nom
‘games’
Agentive Nominalization
of Class 11
u-pepe-aji
cl11-fan-agent_nom
‘fanning, waving’
Instrumentive
Nominalization of Class 11
u-pep-o
cl11-fan-inst_nom
‘wind’
Agentive Nominalization
of Class 1/2
m-pig-an-aji
cl1-hit-recip-agent_nom
‘fighter’
wa-pig-an-aji
cl2-hit-recip-agent_nom
‘fighters’
Instrumentive
Nominalization of Class
3/4
m-pig-an-o
cl3-hit-recip-inst_nom
‘battle’
mi-pig-an-o
cl4-hit-recip-inst_nom
‘battles’
Agentive Nominalization
of Class 11
u-pig-an-aji
cl11-hit-recip-agent_nom
‘rivalry’
Instrumentive
Nominalization of Class 11
u-pig-an-o
cl11-hit-recip-inst_nom
‘contest’
Complex Verb
piga-na
‘hit-recip’
piga-na
‘hit-recip’
179
Nick Kloehn
complex form (happiness), and does not occur when the ratio is the reverse. The model
encompassing these concepts has been termed ‘The Dual Route Model.’ This tendency to
parse has been argued to be hierarchically ordered from more separable units, which can
freely attach to a base or stem (-ness), to less separable affixes which may only attach
to a base (-th) (Hay & Plag 2004, Plag & Baayen 2009) pending semantic or syntactic
selectional restrictions.
However, subsequent research has argued against this model by suggesting that morphological decomposition must occur in all cases (Taft 2004), that an information theoretical analysis will outperform the frequency accounts described above in predicting
reaction times in a Visual Lexical Decision task (del Prado Martín et al. 2004) and that
complex word-form frequency is a better predictor of decomposition overall (Baayen et
al. 2007). So far, this research suggests that words are stored near their morphological
neighbors, and are arguably stored in either morphologically complex or simplex forms.
1.4 Morphological productivity
The aim of this section is to describe the line that has been drawn from the status multimorphemic word storage (as one unit or as multiple units) to claims about morpheme productivity. Morpheme productivity is defined as the ability of a morpheme to occur in new
environments, or to occur with novel words. For example, in English re- is qualitatively
more productive than en-, because it can occur with words new to the language more
readily (e.g. re-google: to google again versus #engoogle). Early research questioning the
status of productivity has been qualitative (Schultink 1961) or as extremely difficult to
measure (Aronoff 1976). However, newer models have attempted to categorize productivity as a consequence of word frequency. Bybee (1995) claims that token frequency (e.g.
number of occurrences of re+visit) and type frequency (e.g. number of occurrences of
re+STEM ) are indicative of productivity. From this perspective, when a complex wordform has an individual token frequency that is relatively high, the word is more likely
to be stored as a whole unit in one’s mental lexicon and thus may not be decomposed.
Words with a relatively low token frequency will tend to be novel, and if they are associated with a high type frequency then all of these words will be stored by each other in
the lexicon. This storage, she argues, creates a stronger representation of the affix (e.g.
re-). Bybee asserts that a stronger representation for a given morpheme should generally be associated with an increase in its productivity. Whereas Bybee’s analysis does
not make claims about a numeric threshold for this effect, Alegre & Gordon (1999) argue
for a numeric threshold for a token frequency effect for complex forms above six per
one million.
Baayen (1992) and Hay & Baayen (2002) claim that comparing token frequency (e.g.
re+visit) and type frequency (e.g. re+STEM) alone is insufficient to account for productivity. Rather, the number of different decomposed forms acts as the predictor of productivity. They define decomposition as a function of the relationship between the derived
form of a complex word (e.g. re+visit) and its underived or base (e.g. visit). If the derived
form is lower than the base form, then the derived form will be parsed in perception. This
180
11 Towards a unified theory of morphological productivity in the Bantu languages
is measured by looking at the underived versus derived frequency of forms containing
the affix in question (Hay 2002). If enough of these forms that contain the given affix
(e.g. all words that contain the affix re-) are parsed in perception, then the morpheme
will have an autonomous representation in the lexicon, and will therefore be productive.
Here, there is a direct line between parsing and production: if an affix is parsed more
often than memorized, it is stored separate from the words it affixes to, and is therefore
predicted to be productive.
1.5 Rationale for present study
The research described has evaluated the relationship between frequency and productivity in languages like Dutch, English, French and Spanish (Baayen 1992, Bybee 1997, and
Hay & Baayen 2002) but little work of this form has been conducted in more synthetic
languages like Swahili. One such example is a study in Finnish which found that words
which have larger numbers of morphologically related words elicit correct responses in
a visual lexical decision task faster than those who had have smaller number of related
words (del Prado Martín et al. 2004). The researchers claim that when a word in Finnish
is morphologically related to many derived or complex words, those words tend to be
accessed faster as opposed to the overall number of morphologically similar words in
the language as is the case in Dutch. In another study on Finnish, Lehtonen et al. (2006)
used fMRI to test neural activation during the recognition of complex words. They found
that words which were morphologically complex tended to elicit activation in the region
associated with semantic and syntactic processing, and that this might be evidence for
morphological decomposition in perception.
Besides Finnish, other research has been done on languages with higher degrees of
synthesis than previous languages including work on the productivity of nominal morphemes in Japanese. One such example is a study which tested both aphasic and nonaphasic speakers of Japanese on how they processed two nominal suffixes which varied
in their productivity (Hagiwara et al. 1999). They claim that most adjectives in Japanese
may derive a nominal by co-occurring with the highly productive -sa (e.g. takai → takasa
versus kebai → kebasa), but only a subset of adjectives may derive a nominal by cooccurring with the less productive -mi (e.g. takai → takami versus kebai → *kebami).
They found that the processing of the two suffixes coincided in the activation of two
areas: Broca’s area for the productive (-sa) morpheme and the left-middle and inferior
temporal areas for the semi-productive morpheme (-mi). To them, the difference in activation provides evidence for the claim that these two morphemes vary in a qualitative
degree of productivity.
Overall, the work on typologically synthetic languages has suggested a difference in
how these languages are processed compared to less synthetic languages. Whereas the
results in the Japanese seem to support the predictions made by the DRM, nothing has
been down to evaluate productivity in a language with the degree of synthesis in Swahili.
The only experiment on Lexical Access in Swahili has argued that morphologically related words are stored near each other in the lexicon for second language speakers. Foote
181
Nick Kloehn
et al. (2014) performed an experiment in Visual Masked Priming aimed at asking whether
one can prime noun class prefixes within a semantic class (e.g. does m-tu ‘cl1.sg-man’
prime wa-tu ‘cl2.pl-man’?) for L2 Swahili speakers. They found that when the primes
were related by noun class (e.g. vi-tanda ‘cl8.pl-bed’), there was an accelerated reaction
time in accurately identifying the target word (e.g. kitanda ‘cl.7-sg-bed’), but not to the
same extent as to when the prime and target were identical (e.g. kitanda - kitanda). This
suggests that words in the same noun class are stored in the same area in the mental lexicon for L2 speakers of Swahili. However, these findings are only a small step towards
evaluating the relationship between productivity and word storage in Swahili. Therefore,
this study evaluates the relative frequencies of nominal forms as outlined by the DRM in
order to evaluate whether any asymmetry in productivity is predicted. This is performed
by investigating the agent nominalizing affix -aji and the degree to which agentive nominalization interacts with the noun classification and verbal extension system. In order
to see how often the suffix occurs (token frequency) and how many different word forms
contain the suffix (type frequency) a frequency list is created from a corpus representing a sample population of Swahili words. In addition, token frequency is compared to
type frequency and hypotheses are made to the likeliness of forms to be decomposed,
and therefore the degree to which -aji may or may not be productive. Next, the log frequency of every instance of the morpheme (token derived frequency) is compared to the
log underived frequency of each token in the same corpus (token underived frequency)
in order to look at the overall trend in aggregate frequencies. Lastly, the number of morphemes per word is compared to the degree of predicted productivity of individual forms
to determine whether a higher number of morphemes is correlated to a higher degree
of predicted productivity.
2 Evaluating productivity
The aim of the current section is to describe a corpus analysis investigate that words that
occur with the agentive nominalizing suffix -aji, and to answer whether the process can
be described as productive based upon the models described above (Bybee 1995, Hay &
Baayen 2002). The corpus study was performed using the the Helsinki Swahili Corpus
of approximately 12.6 million words (HCS 2004) compiled from 12 news sources. The
corpus of modern Swahili was used as a sample of the synchronic language present
in speakers minds such that frequencies of word forms listed may very well reflect the
frequency of occurrence of words contained in a speaker’s lexicon. This commonly made
behavioral inference leads us to the idea that frequency reflects the degree to which
representations of a given word form has been accessed, and therefore will lead us to
the idea ultimately of how productive these forms may be. In order to get around the
issue as to whether complex verbs could be considered underived forms, all forms were
included and controlled for in both analyses.7
7 This
issue could be a paper in itself. Hay & Baayen (2002) consciously limit their analysis to root forms
and root+1 morpheme forms exactly because multiple affixation conflates the effect of relative frequency.
Essentially, measuring the effect of underived versus derived frequencies get complicated when you have
multiple affixation which themselves can denote both a derived and underived form relative to each other.
182
11 Towards a unified theory of morphological productivity in the Bantu languages
2.1 Measure 1: Token versus type frequency
The current subsection describes a Token versus Type frequency analysis of agentive
deverbal nouns in Swahili. This was performed by using a regular expression to acquire
all forms that ended in -aji along with the token frequency of each form. The regular
expression used to obtain these words was unrestricted in order to include every possible
token occurrence, and thus the resulting list required editing in order to filter out words
that contained the word sequence -aji but were not related to the derivational suffix (e.g.
haji ‘pilgrim’; jaji ‘judge’) as well as words that were misspelled (e.g. *nitakusemaji cf.
nitakusemaje ‘how I will tell you’). In order to obtain the type frequency (all occurrences
of -aji), the token frequencies for each of the findings were summed. Each item was
manually coded for whether it was simplex of complex,8 and for its noun class. The
resulting list was then analyzed in R, and a frequency table was made combining Noun
Classes which vary only in number. Table 2 depicts the finding per Noun Class.
The data depicted in Table 2 demonstrate multiple important aspects of agentive nominalization in Swahili. First, there is a high tendency for nominalized forms to occur in
2 semantic fields (i.e. class 1/2 and class 11). Since class 1/2 denotes animates, and class
11 denotes abstract concepts the data show that there is a strong association of -aji with
animacy, and with the abstract class. Second, the Token Frequency for each Noun Class
(save for Class 5/6) is above the 6 per million threshold argued by Alegre & Gordon (1999)
to be the minimum frequency for the parsing of complex forms. This indicated that for
most of these forms, we would predict that the frequencies are not too low to exhibit
8 Complex
is divided into two groups: 1 verbal extension present or 2 verbal extensions present.
Table 2: Agentive deverbal nouns. Number of occurrences per million of tokens of the type STEM-aji in the Helsinki Swahili Corpus of 12,610,158 tokens.
These tokens are divided by noun class ranging from the highest frequency to
the lowest frequency occurrences per class. The Type Frequency column depicts how many unique words occur in each class per million, and the Token
Frequency column depicts the raw number of words occurring in each class
per million (given that many words have repeat occurrences this number is
higher than the type frequency). The percentages highlight the proportion of
total tokens that occur in a given noun class. The third columns shows that
ratio of Type to Token occurrences per noun class.
Noun class
Type frequency
(per million)
m-/wa- (1/2)
u- (11/14)
ki-/vi- (7/8)
ji-/ma- (5/6)
n-/n- (9/10)
24.11
23.39
1.98
0.95
0.56
Total
50.99
47.28%
45.87%
3.88%
1.86%
1.11%
Token frequency
(per million)
Type:Token
876.03
611.80
29.90
4.20
6.26
0.0275
0.0382
0.0662
0.2262
0.0895
57.32%
40.04%
1.96%
0.27%
0.41%
1528.19
183
Nick Kloehn
parsing in perception. Lastly, and most relevant to productivity: the ratio of token and
type frequencies are very similar between noun classes. A fair assertion to make here is
that, given that the ratios are not wildly different, we would predict of similar degree of
productivity across the classes if productivity is in fact a function of Type versus Token
frequency. This would indicate that the tendency for tokens to occur in only two classes
is a likely a semantic effect and not an effect of productivity.9
In order to determine whether productivity is predicted to occur equally in each noun
class, a one-way ANOVA was used to test whether there has a correlation between noun
class and the Type: Token ratio. Significance would suggest that if productivity were
truly a function of Type:Token ratio, then we would predict asymmetries in productivity
between noun classes, and non-significance would predict no productivity asymmetries
between noun classes. Type:Token ratio did not significantly vary across the noun classes
F(1,574) = .767, p < .35. This may indicate that agentive deverbal nominals will be equally
productive across the noun classes, and that asymmetry in the overall frequency of these
classes is conditioned by semantic restriction.
Collapsing across the classes, we can assess productivity from the perspective of Token versus Type frequencies by identifying the average number of occurrences for each
type. According to Bybee, a productive morpheme would have many tokens that occur
only a few times, as opposed to a few forms that occur many times. Figure 1 represents
the token frequency for each Type as a histogram.
Figure 1: Histogram depicting the log of the Token Frequency along the x-axis,
and the Frequency of Types occurring per log token frequency on the y-axis.
This means that the bar furthest left depicts a group in which there are 250
tokens with a log frequency of zero (i.e. a frequency of 1, given that Log(1)=0 ).
9
This notion stems from the idea that these noun classes are both semantically and functionally determined.
Whereas class 1/2 and 11 have strong semantic distinctions, many others do not indicate semantic fields that
are as consistently clear. In addition, most new words are put into class 9/10 (often unmarked) indicating
a non-semantic classification, based on grammatical necessity of noun class membership in the language.
184
11 Towards a unified theory of morphological productivity in the Bantu languages
The histogram depicts data much like what Bybee’s productive morpheme would resemble. There are around 250 type types with a log frequency occurrence of 0, moving
right, there are over 400 types with a Log Token Frequency of less than 2, around 520
less than 4, and possibly only 50 types greater than 4. The result is clear: the majority
of types have a relatively low token frequency, and only a few outliers represent the
bulk of token frequency (more than 10 tokens).10 The data show that one would predict
productivity from the standpoint of a model that draws upon productivity from Token
versus Type frequency analysis. If the relationship between Type and Token frequency is
a cue for productivity, and not overall raw frequency, then one would predict a speaker
to be equally accepting of a novel form in one noun class as opposed to another barring
semantic ill-formedness.
2.2 Measure 2: Underived versus derived frequency
The aim of this subsection is to build on the analysis in the previous section in making
a case for productivity by testing agentive deverbal nouns for the derived versus underived frequency. This was performed by using the list of all tokens from the previous
analysis. A Perl regular expression was then used to strip all deverbal nouns of their
noun class prefixes and agentive suffixes in order to find the underived forms. Given
that verbs in Swahili may not occur without left peripheral inflectional prefixation, the
underived words included verbs containing inflectional affixation.11 No complete form
of the Helsinki Swahili Corpus is openly available, and so frequency information may
only be obtained by web query. Given the number of underived forms tested (574), the
process was automated using a BASH cURL script which saved the returned token and
frequency information in JAVA script. The results were concatenated into a single text
file, and search and replace commands were used to transform the data into analyzable
columns, to then be added to the data frame from the previous analysis.
In the vein Hay & Baayen (2002), the log frequencies of the underived (base) tokens
were graphed against the log frequency of the derived tokens. Figure 2 demonstrates the
relationship between base frequency (underived form) and derived frequency of each
occurrence of -aji within the corpus. Within the DRM, whether or not a form is parsed
is related to its position on this graph to an X=Y line. This is simply a line with a slope of
1, and a y-intercept of 0 that bisects points where the x and y measure are equivalent. In
addition, they require that an r2 line describing the data should be significantly above the
X=Y line in order for the morpheme to be productive. According to Hay & Baayen, this
10
It is important to note here that forms that are derivationally related but vary in semantic noun class and
not just number (e.g. word forms that occur in class 1 and 11) have not been combined into the same type
category. An analysis combining the two or more types would gather together words that differ in in one
derivational morpheme, but who share the same derivational root. Whether or not the frequency of one
should effect the other will be left to future studies.
11
Swahili verbs inflect as follows, with [] indicating obligatory inflection, and {} indicating optional inflection/derivation based upon the denotation of the proposition: [Subject]-[Tense/Aspect]-{Object}-Verb
Stem-{Verbal Extension(s)}-Final Vowel Chakula ki-na-ku-pik-i-w-a. cl7.food cl7.agr-pres-2sg.obj-cookappl-pass-fv ‘The food is being cooked for you.’
185
Nick Kloehn
relationship reveals whether we can predict that the morpheme in question has some
sort of mental representation, and therefore occurs productively as seen in Figure 2.
Figure 2: Log Derived Frequency pitted against Log Base Frequency. The blue
dotted line represents that X=Y line. A data point on this line will have an equivalent derived and underived frequency. Points above this line have a higher Log
Base Frequency than a Log Derived Frequency, and ones below have a higher
Derived Frequency than Base Frequency. The green solid line represents the
r2 line (p > .0001) which demonstrates the significant positive correlation between Base Frequency and Derived Frequency.
The Figure above shows data points that relate a single forms derived and underived
frequencies. A data point below the line will not be parsed in perception according to
the theory, and one above the line would be parsed. Clearly the bulk of the data is above
the X=Y which means we would predict parsing in perception. The r2 line describes this
phenomenon mathematically. The relationship between the X=Y line and the r2 line is
significantly above the X=Y line, we can posit that the bulk of data points have a higher
underived frequency than derived frequency. It is precisely this relationship that denotes
productivity in the Dual Route Model.
The aim of this section was to strengthen a claim for productivity of the agentive
nominal from the perspective of a dual processing route model. Both theories predict
productivity of the affix. The next section aims to expand on this claim by analyzing
morphological complexity within the nominal forms.
3 Morphological complexity and productivity
The aim of the current section is to analyze the relationship of the verbal extension system to the degree of how productive a given form is. We would predict that the presence
of verbal extensions impacts productivity, then there will be an asymmetry in the predicted productivity of verbs containing verbal extension and those which do not. To do
186
11 Towards a unified theory of morphological productivity in the Bantu languages
this, the residuals of the r2 line from §2 (the distance from the line describing the distribution) are compared to the number of verbal extensions present in the forms.
3.1 Regression analysis
In order to test this relationship, morphological complexity was tested against the degree morphological productivity measure using the residuals. A residual of zero means
that the data point is on the r2 line, and points above and below are indicated by positive and negative integers respectively. The assertion here is that points with a negative
residual, while still being parsed, are working against productivity, whereas a positive
residual is working towards a higher level of productivity. The question then is, is there
any correlation between this tendency and whether or not there is a verbal extension
present?
3.2 Results
A one-way ANOVA was used to test whether morphological complexity influenced distance for the r2 line. Given the unevenness of scores per group (300 in Simplex, 150 in 1
Complex, and 59 in 2 Complex) a random sample of data was taken from the two larger
groups of equal number to the smallest group. The groups Distance from the r2 line do
not vary significantly across the three groups, F (1, 58) = 26.89, p > .2. However, as the
box plot in Figure 3 shows, the non-truncated data shows a trend toward significance in
which there is an inverse relationship between morphological complexity and distance
from the r2 line. However, a nonsignificant affect suggests that the predicted productivity of denominal verbs is not impacted by whether or not verbal extensions are present.
Figure 3: Box plot showing the three groups of morphological complexity on
the x-axis (0 Verbal Extensions, 1 Verbal Extension, and 2 Verbal Extensions),
and the distance to the regression line on the y-axis. The medians of the groups
nonsignificantly decrease as the number of verbal extensions increase.
187
Nick Kloehn
4 General discussion
Under each analysis given, one would predict that the affix -aji should be productive for
both complex and simplex verb forms, and regardless of noun classification. The next
question to be answered is whether or not all nominalized forms exhibit the same variation. For example, does the instrumental form -o exhibit the same degree of predicted
productivity with no inhibition for the other morphological systems? A related question
would be to isolate whether there are any morphological processes in Swahili that are
non-productive. A corpus study on a much larger scale will be able to evaluate this by
compare derived and underived forms of all verbs and nominals. If it is the case that
these affixes are all predicted to be productive, then there is a testable prediction for
a behavioral study investigating the ways in which Swahili speakers process complex
words. Namely, the DRM would predict that all forms should be decomposed and therefore that there should be no significant asymmetry in reaction times between complex
word forms beyond their base frequencies. The opposite result would indicate that speakers of Swahili may have different thresholds for parsing than speakers of more isolating
languages like English. Such a finding would provide evidence for the idea that Lexical
Access is ordered relative to the language you acquire. This would entail that some languages would make use of one route of processing more commonly than a language like
English, or that these models are wide of the mark, demanding a reanalysis of the way
in which human language is processed cross-linguistically.
Acknowledgements
The data in this paper were presented at The University of Arizona Graduate Student
Showcase Feb 2014, and ACAL 45 at the University of Kansas. Thanks to all involved and
to Mike Hammond, Heidi Harley, Adam Ussishkin and Andy Wedel for comments.
Abbreviations
Numbers in glosses indicate noun class prefixes and pre-prefixes. Abbreviations follow
Leipzig glossing conventions, with the following exceptions:
agent_nom
aug
agentive nominalizer
augment
fv
stat
final vowel
stative
References
Alegre, Maria & Peter Gordon. 1999. Frequency effects and the representational status of
regular inflections. Journal of Memory and Language 40(1). 41–61.
Aronoff, Mark. 1976. Word formation in generative grammar. Linguistic Inquiry Monographs 1. 1–134.
188
11 Towards a unified theory of morphological productivity in the Bantu languages
Baayen, Harald R. 1992. Quantitative aspects of morphological productivity. In Geert
Booij & Jaap van Marle (eds.), Yearbook of morphology 1991, 109–149. Netherlands:
Springer.
Baayen, R. Harald, Lee H. Wurm & Joanna Aycock. 2007. Lexical dynamics for lowfrequency complex words: A regression study across tasks and modalities. The Mental
Lexicon 2(3). 419–463.
Bertram, Raymond, R. Harald Baayen & Robert Schreuder. 2000. Effects of family size
for complex words. Journal of memory and language 42(3). 390–405.
Broadbent, Donald E. 1967. Word-frequency effect and response bias. Psychological review
74(1). 1.
Bybee, Joan L. 1995. Regular morphology and the lexicon. Language and Cognitive Processes 10(5). 425–455.
Bybee, Joan L. 1997. Semantic aspects of morphological typology. In Joan L. Bybee, John
Haiman & Sandra A. Thompson (eds.), Essays on language function and language type:
Dedicated to t. Givón, 25–37. Amsterdam: John Benjamins Publishing.
De Jong IV, Nivja H., Robert Schreuder & Harald R. Baayen. 2000. The morphological
family size effect and morphology. Language and Cognitive Processes 15(4-5). 329–365.
del Prado Martín, Fermín Moscoso, Aleksandar Kostić & Harald R. Baayen. 2004. Putting
the bits together: An information theoretical perspective on morphological processing.
Cognition 94(1). 1–18.
Foote, Rebecca, Patti Spinner & Rose Mwasekaga. 2014. Morphological decomposition in
Swahili. Presentation at Linguistic Society of America 2014 Annual Meeting.
Greenberg, Joseph H. 1959. Africa as a linguistic area. In William R. Bascom & Melville J.
Herskovits (eds.), Continuity and change in African cultures, 15–27. Chicago: University
of Chicago Press.
Hagiwara, Hiroko, Yoko Sugioka, Takane Ito, Mitsuru Kawamura & Jun I. Shiota. 1999.
Neurolinguistic evidence for rule-based nominal suffixation. Language 75(4). 739–763.
Hay, Jennifer. 2002. From speech perception to morphology: Affix ordering revisited.
Language 78(3). 527–555.
Hay, Jennifer & Harald R. Baayen. 2002. Parsing and productivity. In Geert Booij & Jaap
van Marle (eds.), Yearbook of morphology 2001, 203–235. Netherlands: Springer.
Hay, Jennifer & Ingo Plag. 2004. What constrains possible suffix combinations? On the
interaction of grammatical and processing restrictions in derivational morphology.
Natural Language and Linguistic Theory 22(3). 565–596.
Heine, Bernd. 1982. African noun class systems. Apprehension: das sprachliche Erfassen
von Gegenstanden 1. 189–216.
Katamba, Francis X. 2003. Bantu nominal morphology. In Derek Nurse & Gerard Phillippson (eds.), The Bantu languages, 103–120. London: Routledge.
Lehtonen, Minna, Victor A. Vorobyev, Kenneth Hugdahl, Terhi Tuokkola & Matti Laine.
2006. Neural correlates of morphological decomposition in a morphologically rich
language: An fMRI study. Brain and Language 98(2). 182–193.
Mohamed, Mohamed Abdulla. 2001. Modern Swahili grammar. Nairobi: East African Publishers.
189
Nick Kloehn
Plag, Ingo & Harald R. Baayen. 2009. Suffix ordering and morphological processing. Language 85(1). 109–152.
Schultink, Henk. 1961. Produktiviteit als morfologisch fenomeen. Forum der letteren 2.
110–125.
Seidl, Amanda & Alexis Dimitriadis. 2003. Statives and reciprocal morphology in Swahili.
Typologie des langues d’Afrique et universaux de la grammaire 1. 239–284.
Taft, Marcus. 1979. Recognition of affixed words and the word frequency effect. Memory
& Cognition 7(4). 263–272.
Taft, Marcus. 2004. Morphological decomposition and the reverse base frequency effect.
Quarterly Journal of Experimental Psychology Section A 57(4). 745–765.
TUKI, The Institute of Swahili Research at the University of Dar es Salaam. 2001. English
- swahili dictionary - kamusi ya kiingerza-kiswahili. Dar es Salaam: Laurier Books Ltd.
ISBN 9976911297.
190
Chapter 12
The acoustic vowel space of Anyi in
light of the cardinal vowel system and
the Dispersion Focalization Theory
Ettien Koffi
St. Cloud State University
The Cardinal Vowel System (CVS) and the Dispersion Focalization Theory (DFT) make an
important assumption about the inventory of vowels in world languages. The claim is that
languages organize their vowels in a certain way in the auditory-perceptual space so as
to maximize intelligibility. The vowel diagrams of African languages in influential publications such as Welmers (1973: 20–45) explicitly or implicitly reflect this assumption. However,
persistent confusions between [ɪ] and [e] among Anyi Morofu speakers have aroused my
curiosity and led me to investigate the matter acoustically. The findings reported here show
that the vowel space of Anyi Morofu is in a between and betwixt state. The data indicates
that this dialect is moving from a nine-vowel system to an eight-vowel system through the
merger of [ɪ] and [e]. There are also signs of the impending merger of [ʊ] and [o].
1 Introduction
The Cardinal Vowel System (CVS) and the Dispersion Focalization Theory (DFT) agree
on the principle that languages organize their vowel inventories in order to maximize
intelligibility. The principle underlying both approaches is known as the Principle of
Perceptual Separation (PPS). Ladefoged & Johnson (2015: 238) explain it as follows, “One
of the forces acting on languages may be called the principle of sufficient perceptual
separation, whereby the sounds of a language are kept acoustically distinct to make it
easier for the listener to distinguish one from another.” This important principle collides
with how Anyi Morofu organizes its vowel inventory. Anyi Morofu is the biggest dialect
of the Anyi language spoken in Côte d’Ivoire, West Africa. According to the 2000 census
(outdated, but there is no other official census data to go by), this dialect is spoken by
more than half of the 755,365 Anyi speakers in Côte d’Ivoire. Anyi belongs to the Akan
family of languages. Before presenting the evidence for how Anyi Morofu runs counter
to the core principle of CVS and DFT, let’s acquaint ourselves briefly with these two
phonetic frameworks. The goal here is not to review these two theories extensively, but
Ettien Koffi. The acoustic vowel space of Anyi in light of the cardinal vowel system and the
Dispersion Focalization Theory. In Jason Kandybowicz, Travis Major, Harold Torrence & Philip
T. Duncan (eds.), African linguistics on the prairie: Selected papers from the 45th Annual Conference
on African Linguistics, 191–204. Berlin: Language Science Press. DOI:10.5281/zenodo.1251730
Ettien Koffi
rather to use the data they provide to explain the perceptual confusion between [ɪ] and
[e] and the signs announcing the upcoming merger of [ʊ] and [o].
1.1 A short history of the cardinal vowel system
On the occasion of Ladefoged’s sixtieth birthday, Fromkin (1985) put together a collection
of papers from influential phoneticians. In Abercrombie’s article (1985:17), he gives us a
glimpse of how CVS came about. His account is authoritative because he had a front
row seat when Jones was designing his method. He was Jones’ student and later became
Ladefoged’s teacher and mentor. He notes that for Jones, CVS was not a theory, but a
technique. He describes this technique as follows:
This way of teaching phonetics meant intensive training of the proprioceptive, i.e.,
the tactile and kinesthetic senses concerned with the organs of speech, something
that is not valued very highly by many other schools of phonetics. The proprioceptive senses, in the view of phoneticians in the Jones tradition, play an important
part in the analysis and description of unfamiliar sounds. The phonetician, having
learnt to make a sound of the language he is working on to the complete satisfaction of his native informant, then examines what he himself is doing with his vocal
organs, and infers the informant is doing the same.
Jones learned to produce a wide variety of vowels this way. Thomas (2011: 146) provides in Table 1 the formant frequencies of 18 vowels that Jones learned to produce.
For the purposes of this paper, the focus will be on nine vowels, [i, ɨ, e, ɛ, a, ɔ, o, ʉ,
u], because Anyi also has nine vowels. According to Maddieson’s (1984) UCLA Phonetic
Segment Inventory Database (UPSID), 17 out the 266 languages in the database have
nine vowels. Languages such as Anyi with a nine-vowel system represent only 6.39% of
the total number of languages in UPSID. Furthermore, only seven of the 17 nine-vowel
languages have a perfect symmetry of four front vowels and four back vowels and one
low central vowel. The vowel system of these seven languages is similar to the one we
find in Anyi. Jones’ cardinal vowel system did not include [ɪ] and [ʊ] because [-ATR]
vowels were not known at the time. Even so, the plotting of his vowels gives us a realistic
picture of what a nine-vowel system looks like.1
A few cursory remarks need to be made. First and foremost, in the nine-vowel system
produced by Jones, we see that the PPS obtains. No two vowels overlap in acoustic space.
CVS has had a far-reaching impact on how the vowel inventories of African languages
are plotted in Welmers (1973: 20–45), in Atlas des Langues Kwa de Côte d’Ivoire, Tome
1, and in countless other publications. Thomas (2011: 145–147) opines that Jones’ original intention in proposing CVS was only to “standardize impressionistic transcription
to make it more useful for interlanguage comparisons,” not to idealize it as the acoustic
vowel spaces for all languages. Koffi (2009) and all who have described the Anyi vowel
quadrant have used this idealized system. This is the reason why the confusion between
[ɪ] and [e] came as a surprise because under the idealized Anyi vowel quadrant, unintelligibility was not expected.
1
192
The diagrams were produced using Norm, available at http://lingtools.uoregon.edu/norm/norm1.php.
12 The acoustic vowel space of Anyi
Table 1: Jones’ Vowels
N0
Vowels
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
[i]
[ɨ]
[ɯ]
[u]
[y]
[ʉ]
[e]
[ø]
[ɤ]
[o]
[ɛ]
[œ]
[ɶ]
[ɒ]
[ʌ]
[ɔ]
[ɑ]
[a]
F1
F2
F3
266
312
337
248
289
285
376
353
569
354
588
554
722
582
542
522
650
929
2581
2078
1275
490
2231
1487
2213
1946
1153
724
1910
1549
1227
769
1145
932
940
1688
3627
2544
2180
2512
2747
2066
2652
2375
2282
2348
2328
2158
2180
2150
2273
2180
2472
2354
Figure 1: The Norm website does not recognize certain IPA symbols. The following legend is used <ii> = [ɨ], <uu> = [ʉ], <ee> = [ɛ], <oo> = [ɔ] instead.
193
Ettien Koffi
1.2 A quick overview of the Dispersion Focalization Theory
PPS is also at the core of the Dispersion Focalization Theory (DFT) that Schwartz et al.
(1997) put forth. However, Becker-Kristal (2010: 10) contends that “the idea that vowel
inventories are structured in a manner that enhances contrast, by maximally dispersing
vowels in the auditory-perceptual space, is as old as the intuition that vowel inventories
follow universal structural patterns.” It is worth stating clearly and unambiguously that
the goal pursued in this paper is not to review, critique, or summarize DFT or the Dispersion Theory (DT) from which it sprang. Such an exercise would require us to make a long
detour in the histories and developments of these two theories. It is not the theoretical
claims of DFT that interest us as much as the impressive amount of formant frequency
data provided for 22 “prototypical” vowels, as seen in Table 2. Aspects of this data will
be used in Figure 2 to highlight the acoustic vowel space of a prototypical language with
a nine-vowel system.
Table 2: Prototypical Vowel Frequencies
N0
Vowels
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
[i]
[ʏ]
[ɨ]
[ɯ]
[u]
[ɪ]
[y]
[ʊ]
[e]
[ø]
[ə]
[ɤ]
[o]
[ɛ]
[œ]
[ɜ]
[ʌ]
[ɔ]
[æ]
[ɐ]
[ɑ]
[a]
F1
F2
F3
277
277
277
277
277
344
344
344
414
414
414
414
414
565
565
565
565
565
648
648
735
800
2208
1937
1520
1218
553
2170
1770
635
2065
1608
1516
1248
721
1819
1520
1462
1258
915
1712
1405
1278
1228
3079
2232
2310
2500
2420
2660
2230
2413
2570
2250
2500
2500
2406
2528
2500
2500
2500
2373
2490
2500
2500
2500
Nine of the prototypical vowels, [i, ɪ, e, ɛ, a, ɔ, o, ʊ, u], are also found in Anyi. They are
plotted in Figure 2 to show how these prototypical vowels are organized in an acoustic
space.
194
12 The acoustic vowel space of Anyi
Figure 2: Vowel space of nine prototypical vowels
The Norm website does not recognize certain IPA symbols. The following legend is used for [-ATR] vowels: <ii> = [ɪ], <uu> = [ʊ], <ee> = [ɛ], <oo> = [ɔ].
The plotting shows that the claims of PPS hold here as they did in Figure 1. No vowel
encroaches on the space of another vowel. Consequently, intelligibility is maximized.
Let’s now turn to the Anyi Morofu data and examine its vowel space in light of CVS and
DFT.
2 Data collection and participants
Koffi (2009), Quaireau (1987: 27), Retord (1980: 96), to name only the three main researchers on Anyi, have all diagrammed the oral vowels of Anyi as shown in Figure 3:
High Front
u
i
ʊ
ɪ
Mid
e
o
ε
Low
High Back
ɔ
a
Figure 3: Anyi Morofu vowel diagram
Anyi also has seven nasal vowels: [ɪ, ĩ, ũ, ʊ, ɛ, ɔ, ã]. The vowels [ɛ] and [ɔ] are deemed
unnasalizable in some Akan languages, but not in Morofu. The only vowels that are unnasalizable are [e] and [o] (Koffi 2004). Figure 3 may be an accurate representation of a
195
Ettien Koffi
nine-vowel phonemic system in Anyi Morofu, but it is no longer an accurate representation of its contemporary phonetic vowel system. This became abundantly clear during
a literacy seminar in the summer of 2011. During the dictation task, teachers in training
confused [ɪ] with [e], and [e] with [ɪ] regardless of who was doing the dictation.2 Test
takers would frequently stop to ask the reader whether he meant [e] or [ɪ] in instances
where the contextual cues were not enough to disambiguate the lexical items containing
these vowels. For example in a sentence such as, ɔ’a hɪ nnaán ‘he trapped an animal’,
some test takers wrote ɔ’a he nnaán ‘he shared/gave away some of his meat’. The high
number of confusion incidences such as these caused me to wonder if a merger was
happening between these two vowels in the Morofu dialect spoken in the Bongouanou
area. Figure 4 shows this dialect in relation to the other Anyi dialects. As noted earlier,
Morofu has more speakers than all the other dialects of Anyi combined.
Figure 4: The Anyi dialect area
The matter was investigated further through data collection in the summers of 2012
and 2013 after securing the approval of the Institutional Review Board (IRB) from my
university. The same ten male adult literacy teachers were invited again. Female speakers
were not intentionally excluded. At that time, there were no female literacy teachers. The
situation has now changed and we have three female teachers. The lack of female data
does not affect the present analysis negatively because most of the various predictions
of DFT and DT are based on male speech (Becker-Kristal, 2010:31). The participants are
all bilingual in Anyi and French. They range in age from the 30s to 50s. Each participant
produced nine sentences, each containing one of the nine vowels under consideration:
(1)
2
a. <ɔ’a hi> (he/she has refused to eat it)
b. <ɔ’a hɪ> (he/she has caught it)
Ladefoged (2003: 126,130–131) tells a similar story about Banawa, a language of the Amazonian rain forest in
Brazil where there was confusion between [u] and [o] that led to strong disagreements in the orthography
of the language.
196
12 The acoustic vowel space of Anyi
c.
d.
e.
f.
g.
h.
i.
<ɔ’a he> (he/she has shared it)
<ɔ’a hɛ> (he/she is late)
<ɔ’a hu> (it has boiled)
<ɔ’a hʊ> (a nonsense word)
<ɔ’a ho> (he/she has dug a whole)
<ɔ’a hɔ> (he/she has left)
<ɔ’a ha> (he/she has bitten)
Each sentence was repeated three times, for a total of 30 repetitions. The data set
consists of 270 items (9 x 3 x 10). The data was collected on an Olympus Digital Voice
Recorder WS-710. The participants wore a Panasonic head-mounted, noise cancellation
fixed microphone. The recording took place in a quiet room on the premises of the Anyi
Literacy and Translation Center (CATA).
2.1 Methodology
The elicitation word in each sentence begins with /h/. These words were chosen intentionally in order to replicate Peterson and Barney’s methodology as much as possible.
Countless studies of vowels have followed this methodology. Ladefoged (1996: 112) explains the benefits of choosing /h/ in these kinds of acoustic phonetic studies as follows:
As the positions of the articulators during the sound [h] are similar to those of the
surrounding sounds, such as the adjacent vowels, the frequency components in [h]
sounds have relative amplitudes similar to those in vowels; but the complex wave
has a smaller amplitude and no fundamental frequency, as it is not generated by
regular pulses from the vocal cords.
Since [h] exists in Anyi as an allophone of /k/, Peterson and Barney’s methodology can
be replicated without any problem. The entire duration of the vowel, from the onset to the
offset, was measured. It was not deemed necessary for this study to take measurements
at various points in the vowel because the environment in which the vowel occurred did
not foster co-articulation. Furthermore, the methodology used by Peterson and Barney
that is being replicated in this study did not sample vowels at multiple intervals. The
onset of each vowel was easily identified because of the frication noise contained in [h].
However, it was more challenging to determine the offset of vowels. In annotating the
offset, Thomas (2011: 142) proposes three options:
… The same problem crops up frequently with vowels before a pause. In these
cases, you have another choice to make. One option is to look for a spot where the
vocal fold vibrations become more or less unrecognizable or start looking more like
staticky patterns of aspiration than the sharper pattern usually evident with vocal
fold vibrations. Often, the best way to determine this spot is by moving the cursor
to different spots and listening; after a certain point, all you hear is aspiration, and
197
Ettien Koffi
that point is where you mark the offset. The other option is to mark the offset at
the end of the recognizable aspiration, though this point may be quite difficult to
define.
For this study, the offset of the vowel was determined by following the second option
in Thomas’ recommendation, that is, demarcating the offset right before the point at
which aspiration is heard. The measurements for one speaker were done manually to
ensure that the offsets of vowels are identified accurately. Once the pattern was well
established, Ryan’s (2005) Grid-maker script for Praat was used to annotate all the vowels
produced by the rest of the speakers. Subsequently, Yoon’s (2008) Stress-analysis script
for Praat was employed to collect all the relevant information displayed in Tables 3 and 4.
Various statistical analyses can be run from the measurements in Tables 3 and 4. However, in this study they are used exclusively for the purpose of generating the acoustic
vowel space in Figure 5 and for explaining why Anyi Morofu hearers have a problem
distinguishing [ɪ] and [e] aurally.
Figure 5: Anyi acoustic vowel space
2.2 The reason for the confusion
Figure 5 shows us visually why Anyi Morofu hearers confuse [ɪ] and [e] aurally. We see
that they overlap in perceptual space. The measurements in Table 3 explain why. These
two vowels mask each other aurally because [ɪ] (399 Hz) and [e] (392 Hz) are separated
by only 7 Hz in the F1 domain. It is a well-known fact that a minimum of 20 Hz is needed
for humans to perceive a difference between two sound segments (Ferrand 2007: 34). It
is also well known that the lowest frequency at which a sound is intelligible on an eightoctave frequency band is 63 Hz. Auditory frequency measuring devices and many audio
applications use this baseline as their reference level (Everest & Pohlmann 2015: 12–16).
For acoustic phonetic analyses, this threshold has been rounded down to 60 Hz to make
198
12 The acoustic vowel space of Anyi
Table 3: F1 Formant for all participants
F1
[i]
[ɪ]
[e]
[ɛ]
[u]
[ʊ]
[o]
[ɔ]
[a]
Speaker 1
Speaker 2
Speaker 3
Speaker 4
Speaker 5
Speaker 6
Speaker 7
Speaker 8
Speaker 9
Speaker 10
325
280
307
291
291
304
338
444
255
654
368
407
435
344
415
402
378
469
360
421
355
408
408
368
361
493
381
449
350
356
556
576
623
573
601
662
536
597
584
583
423
329
345
384
396
342
473
420
365
405
539
595
431
583
510
546
506
491
545
490
481
429
414
429
469
556
544
445
544
468
624
639
653
677
639
662
634
654
628
635
942
885
983
882
980
823
981
940
815
1028
Mean
348
399
392
589
388
523
477
635
925
Table 4: F2 Formant for all participants
F2
[i]
[ɪ]
[e]
[ɛ]
[u]
[ʊ]
[o]
[ɔ]
[a]
Speaker 1
Speaker 2
Speaker 3
Speaker 4
Speaker 5
Speaker 6
Speaker 7
Speaker 8
Speaker 9
Speaker 10
2265
2366
2298
2202
2402
1902
1911
2060
2351
2304
2167
2085
2462
2231
2534
1746
1977
2067
2208
2269
2192
2082
2350
2211
2455
1937
1931
2037
2234
1985
2068
1960
2397
2043
2192
1745
1906
2069
2026
1982
1383
856
781
1216
1105
918
2103?
1655
1371
1960
950
859
716
1151
841
1248
817
1206
1804
2228
1694
839
773
891
1041
1754
1429
1637
1985
1885
896
1047
964
1171
946
1103
829
1053
993
1561
1519
1525
1551
1468
1542
1382
1429
1456
1538
1455
Mean
2206
2174
2141
2038
1249
1182
1392
1056
1486
199
Ettien Koffi
calculations simpler (Fry 1979: 68). Labov et al. (2006: 204–221) use it in Atlas of North
American English (ANAE) to assess dialectal variations. Labov et al. (2013: 43) use it to
assess vowel change in Philadelphia. The same 60 Hz threshold is used here to explain the
confusion between [ɪ] and [e], and vice versa. Though F2 and F3 formants contribute to
the overall perception of vowel quality, the calculations of vowel intelligibility are based
on F1 because it alone contains 80% of the acoustic energy in the vowel (Ladefoged &
Johnson 2015: 207).
The practical steps used to assess vowel intelligibility are as follows. On the F1 frequency band, if the acoustic distance between two contiguous front vowels or two contiguous back vowels is ≥ 60 Hz, the two vowels are perceived as distinct. However, if
their acoustic distance falls between 59 and 21 Hz, intelligibility is compromised. If the
acoustic distance between two vowels is ≤ 20 Hz, it means that a merger has taken place
or is taking place. The reason for this is because human beings cannot perceive frequencies lower than 20 Hz. This is exactly what is going on with Anyi Morofu. Hearers in
general have a hard time distinguishing [ɪ] from [e], and vice versa, because the mean
acoustic distance between them is only 7 Hz. There are, however, small inter-speaker
variations. The segments [ɪ] and [e] produced by Speakers 5, 6, and 10 are intelligible
because the acoustic distances between them are respectively 54 Hz, 91 Hz, and 65 Hz.
However, for seven of the speakers [ɪ] and [e] are aurally indistinguishable. For Speakers
1, 2, 7, 8, and 9 the two vowels mask each other because the acoustic distances between
them are ≤ 20 Hz, as shown in Table 5:
Table 5: Inter-speaker variation
F1
[ɪ]
[e]
Distance
Speaker 1
Speaker 2
Speaker 3
Speaker 4
Speaker 5
Speaker 6
Speaker 7
Speaker 8
Speaker 9
Speaker 10
368
407
435
344
415
402
378
469
360
421
355
408
408
368
361
493
381
449
350
356
13
1
27
24
54
91
3
20
10
65
Mean
Standard Deviation
399
37
392
47
7
28
The situation in Anyi is similar in this respect to the merger between [ɑ] and [ɔ] that
is going on in several dialects of American English. For Central Minnesota English, Koffi
(2013: 5) reports that the merger between [ɑ] (855 Hz) and [ɔ] (851 Hz) is complete in the
speech of female speakers because the acoustic distance between the two vowels is only
4 Hz.
200
12 The acoustic vowel space of Anyi
Cross-linguistically, something is going on between [ɪ] and [e] that deserves further investigation. Ladefoged (1999: 41–42) displays the vowels of a southern California speaker
whose [e] has risen above [ɪ]. Koffi’s (2014:16–17) acoustic phonetic measurements of
Central Minnesota English show that that [e] is higher than [ɪ] in male and female speech.
In male speech, [e] (434 Hz) is higher than [ɪ] (542 Hz) by 108 Hz. In female speech, [e]
(508 Hz) has risen above [ɪ] (573 Hz) by 65 Hz. In these examples, the raising of [e] above
[ɪ] does not result in unintelligibility because the acoustic distance between them is still
higher than the 60 Hz threshold. However, this is not so in the case of Anyi Morofu
where only a mere 7 Hz separate these vowels. In the terminology that Schwartz et al.
(1997) use to describe vowel systems, Anyi is an “atypical” nine-vowel system because it
does not conform to the predicted patterns. Becker-Kristal (2010: 169) explains why:
Across all analyses, inventories with ATR harmony often violate the principles of
dispersion, in formant spans, in even vowel spacing and in phonetic adjustments
in response to structural change. These deviations are understandable if such inventories are not treated as one large system but as two parallel smaller systems.
More acoustic phonetic data such as the one used to describe the confusion between
[ɪ] and [e] is needed from other African languages with [±ATR] vowel systems to see if
Anyi Morofu is really atypical or if this phenomenon is widespread. In the case of Anyi,
it is the vowels [ɪ] and [e]. In other languages, it may be different pairs of vowels.
2.3 The future of the Anyi acoustic vowel space
What does the future hold for the phonemic inventory of Anyi vowels? How long before the acoustic vowel space is completely reduced to an eight-vowel system? Will the
acoustic vowel space be reduced further to a seven-vowel system? Becker-Kristal (2010:
113) discusses a possible scenario that may be in store for Anyi:
They [vowels] might fall closer to other vowels, which are repelled further, albeit
by a smaller magnitude, and this process propagates as a push chain shift with
gradual decay through other vowels until the entire system finds a new balance.
What will the new balance look like for Anyi? It is hard to predict the future. However,
we can anticipate what the Anyi Morofu vowel space will look like in the near future by
learning from the current state of vowels in some languages in the Akan family. Mensah
(1983: 430) reports that Krobou, another Akan language, has reduced the number of its
vowels from nine to eight. It no longer has the vowel [ɪ], which has been replaced by
[e]. If Anyi finds a new balance in an eight-vowel system, this balance will be temporary
because another shift is afoot. The data in Table 6 shows that the next vowel targeted for
disappearance is [ʊ] (523Hz):
It will most likely be replaced by [o] (477 Hz). The acoustic distance between them is 46
Hz. This merger may take a little while, but it is inevitable. Only Speakers 2 and 4 mark a
clear contrast between these two vowels. Intelligibility is compromised in the speech of
Speakers 1, 5, and 8. A merger has already taken place in the pronunciation of Speakers 3,
201
Ettien Koffi
Table 6: The Impending Merger of [ʊ] and [o]
F1
[ʊ]
[o]
Distance
Speaker 1
Speaker 2
Speaker 3
Speaker 4
Speaker 5
Speaker 6
Speaker 7
Speaker 8
Speaker 9
Speaker 10
539
595
431
583
510
546
506
491
545
490
481
429
414
429
469
556
544
445
544
468
58
166
17
154
41
10
38
46
1
22
Mean
Standard Deviation
523
48
477
52
46
57
6, and 9; and it is on the verge of happening for Speaker 10. The merger between [ʊ] and
[o] has already taken place in Baule, which is closely related to Anyi. Kouadio (1983: 284)
reports that Baule no longer has [ɪ] or [ʊ]. Other languages in the Akan family spoken
in Côte d’Ivoire have a seven-vowel system instead of the nine traditionally associated
with this language family. Hérault (1983: 262) reports that Avikam has lost both [ɪ] and
[ʊ], and so has Ebrie (Bole-Richard 1983: 324). The acoustic vowel space of Anyi Morofu
will achieve stability when the number of its vowels goes from nine to seven.
3 Summary
The vowel spaces of languages are always shifting. English underwent a major shift
between 1400 and 1600 (Fromkin et al. 2014: 342)). This change has been nicknamed
the Great Vowel Shift. Labov et al.’s (2006) voluminous ANAE shows that another shift
known as the Northern Cities Shift is slowly but surely fanning across the Midwest. Since
change is a language universal process, one would expect the vowels of Anyi to also shift.
The vowel [e] is masking [ɪ] for now. How long will it take for [ɪ] to be swallowed up by
[e]? It is hard to tell. However, the process that is underway is almost irreversible given
what has taken place in other Akan languages that are closely related to Anyi Morofu.
For now Anyi is following the same path as Krobou. In a not so distant future, the shift
from [ʊ] to [o] will run its course, and Anyi will have seven vowels like Baule and other
Akan languages spoken in Côte d’Ivoire. However, synchronically, Anyi is in a between
and betwixt state which causes it to be atypical, that is, it does not conform to PPS as
predicted by CVS and DFT.
202
12 The acoustic vowel space of Anyi
References
Abercrombie, David. 1985. Daniel Jones’ teaching. In Victoria Fromkin (ed.), Phonetic
linguistics: Essays in honor of Peter Ladefoged, 15–24. Orlando: Academic Press.
Becker-Kristal, Roy. 2010. Acoustic typology of vowel inventory and dispersion theory: Insights from a large cross-linguistic corpus. Los Angeles, CA: UCLA Doctoral dissertation.
Bole-Richard, Remy. 1983. L’ebrie. In Georges Hérault (ed.), Atlas des langues Kwa de Côte
d’Ivoire, vol. 1, 307–357. Abidjan, Côte d’Ivoire: Institut de Linguistique Appliquée.
Everest, Alton F. & Ken C. Pohlmann. 2015. Master handbook of acoustics. 6th edn. New
York: McGraw Hill Education.
Ferrand, Carole T. 2007. Speech science: An integrated approach to theory and clinical practice. 2nd edn. New York: Allyn & Bacon.
Fromkin, Victoria. 1985. Phonetic linguistics: Essays in honor of peter ladefoged. Orlando:
Academic Press.
Fromkin, Victoria, Robert Rodman Rodman & Nina Hyams. 2014. An introduction to language. 10th edn. New York: Wadsworth Cengage Learning.
Fry, Dennis Butler. 1979. The physics of speech. Cambridge: Cambridge University Press.
Hérault, Georges. 1983. L’Avikam. In Georges Hérault (ed.), L’Avikam, vol. 1, 255–276.
Abidjan, Côte d’Ivoire: Institut de Linguistique Appliquée.
Koffi, Ettien N. 2004. Oral vowels in nasalization contexts in Anyi: Implications for orthography. Trends in African Linguistics 6. 163–170.
Koffi, Ettien N. 2009. The interface between phonology and morpho(phono)logy in the
standardization of Anyi orthography. Bloomington, IN: Indiana University. http : / /
orthographyclearinghouse.org/index.php/. Revised doctoral dissertation.
Koffi, Ettien N. 2013. The acoustic vowel space of Central Minnesota English: Focus on
intelligibility. Linguistic Portfolios 2. 2–16.
Koffi, Ettien N. 2014. The acoustic vowel space of Central Minnesota English in light of
the Northern cities shift. Linguistic Portfolios 3. 2–20.
Kouadio, Jérémie N. 1983. Le Baoule. In G. Hérault (ed.), Atlas des langues kwa de côte
d’ivoire tome, vol. 1, 277–306. Abidjan, Côte d’Ivoire: Institut de Linguistique Appliquée.
Labov, William, Sharon Ash & Charles Boberg (eds.). 2006. Atlas of North American English: Phonetics, phonology, and sound change. New York: Mouton de Gruyter.
Labov, William, Ingrid Rosenfelder & Josef Fruehwadl. 2013. One hundred years of sound
change in Philadelphia. Language 48. 30–65.
Ladefoged, Peter. 1996. Elements of acoustic phonetics. 2nd edn. Chicago: The University
of Chicago Press.
Ladefoged, Peter. 1999. American english. In Handbook of the international phonetic association: A guide to the use of the international phonetic alphabet, 41–50. New York:
Cambridge University Press.
Ladefoged, Peter. 2003. Phonetic data analysis: An introduction to fieldwork and instrumental techniques. Malden, MA: Blackwell Publishing.
203
Ettien Koffi
Ladefoged, Peter & Keith Johnson. 2015. A course in phonetics. 7th edn. Stamford, CT:
Cengage Learning.
Maddieson, Ian. 1984. Patterns of sounds. Cambridge University Press: New York.
Mensah, Emanuel N. A. 1983. Le Krobou. In Georges Hérault (ed.), Atlas des langues
Kwa de Côte d’Ivoire, vol. 1, 425–463. Abidjan, Côte d’Ivoire: Institut de Linguistique
Appliquée.
Quaireau, André. 1987. Description de l’Agni: Des Parlers Moronou, Ndénié et Bona: Université de Grenoble III Doctoral dissertation.
Retord, Georges L. A. 1980. Etude radiocinématographique des articulations de l’AgniSanvi. Paris Doctoral dissertation.
Ryan, Kevin. 2005. Grid-maker. Praat. no address: no publisher. http : / / phonetics .
linguistics.ucla.edu/facilities/acoustic/praat.html.
Schwartz, Jean-Luc, Louis-Jean Boe, Nathalie Vallée & Christian Abry. 1997. The
dispersion-focalization theory of vowel systems. Journal of Phonetics 25. 255–286.
Thomas, Erik R. 2011. Sociophonetics: An introduction. New York: Palgrave Macmillan.
Welmers, William E. 1973. African Language Structures. Paul F. A. Kotey & Haig DerHoussikian (eds.). Los Angeles, CA: University of California Press.
Yoon, Tae-Jin. 2008. Stress-analysis. Praat. http://web.uvic.ca/~tyoon/resource/vq.praat.
204
Chapter 13
Gender instability in Maay
Mary Paster
Pomona College
This paper discusses variation in the gender of nouns in Maay, a language of Somalia. Languages of the Eastern Omo-Tana subgroup of East Cushitic (including Maay, Somali, Rendille,
and Tunni) have gender systems wherein every noun is masculine or feminine. Masculine
nouns take k-initial variants of suffixes including the definite marker, demonstratives, and
possessive markers; these suffixes are t-initial with feminine nouns. As is now well known,
gender in these languages is sensitive to plurality in various ways: in some languages, gender ‘polarity’ reverses the gender of nouns in the plural; in others, feminine nouns change
to masculine when their plurals are formed with certain suffixes but not others. In Maay,
plurals are all masculine regardless of how they are formed, but the gender of many singular nouns is inconsistent across individuals. The masculine plural pattern makes the gender
of singular nouns unrecoverable from their plurals, so nouns that are frequently plural are
susceptible to gender instability. If there is uncertainty about the gender of some nouns,
speakers may be inclined to guess masculine, thereby producing more feminine to masculine changes than the reverse, due to the prevalence of masculine nouns in the Maay lexicon.
1 Introduction
The Maay language is known to exhibit significant inter-speaker variation in its phonology and morphology (Paster 2013). This paper describes variability in the gender assignment of Maay nouns and considers explanations for why gender is unstable for certain
nouns in this language. I will argue that gender instability is connected to, and facilitated
by, a regular pattern in the language where gender is neutralized to masculine in plural
nouns.
The structure of the paper is as follows. First, in §2 I give some background on the Maay
language and its classification. In §3, I explain the gender neutralization pattern in Maay
plurals and discuss similar phenomena in related languages. §4 describes the problem
of gender instability in Maay. In §5, I propose an explanation of gender instability that
attributes the emergence of gender instability in part to gender neutralization in plurals;
I also consider and reject a number of alternative explanations. §6 concludes the paper.
Mary Paster. Gender instability in Maay. In Jason Kandybowicz, Travis Major, Harold
Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected papers from
the 45th Annual Conference on African Linguistics, 205–218. Berlin: Language Science Press.
DOI:10.5281/zenodo.1251732
Mary Paster
2 Background on Maay
Maay (also known as Af-Maay or MayMay; see Paster 2007; Comfort & Paster 2009;
Paster 2010) is a Cushitic language spoken in Somalia that is related to, but not mutually
intelligible with, Somali. It is classified as an East Cushitic language, for which a tree is
given in (1).
(1)
East Cushitic (modified from Saeed 1999: 3)
East Cushitic
Saho-Afar
Macro-Oromo
Western
Dasenach
Arbore
Omo-Tana
Sidamo
Northern
Elmolo
Burji
Dullay
Yaaku
Eastern
Rendille
Bayso
Boni
Somali
Tunni
Maay
In this paper, I will be focusing on the Eastern Omo-Tana (EOT) subgroup of East
Cushitic, marked in the tree above.
3 Gender in EOT languages
Cushitic studies often refer to the existence of three genders (masculine, feminine, and
plural; cf. Corbett & Hayward 1987). In Maay, this is essentially how third person subject
agreement works for verbs, as seen in four different tenses in (2) (data and tense/aspect
category names are from Paster 2007).
(2)
206
a. Simple Past
roor-i
run-3sgm.past
‘he ran’
b. Simple Present B
ɗeer-ya
be.tall-3sgm.stative
‘he is tall’
roor-ti
run-3sgf.past
‘she ran’
ɗeer-ta
be.tall-3sgf.stative
‘she is tall’
roor-eena
run-3pl.past
‘they ran’
ɗeer-yena
be.tall-3pl.stative
‘they are tall’
13 Gender instability in Maay
c. Immediate Future
kooy-e
come-3sgm.future
‘he will come’
d. Present Progressive
aam-oy-e
eat-pres.prog-3sgm.present
‘he is eating’
kooy-ase
come-3sgf.future
‘she will come’
aam-oy-te
eat-pres.prog-3sgf.present
‘she is eating’
kooy-ayeena
come-3pl.future
‘they will come’
aam-oy-eena
eat-pres.prog-3pl.present
‘they are eating’
For purposes of noun morphology, however, there are only two genders in Maay: masculine and feminine. Masculine nouns (3a) take k-initial suffixes for definites, demonstratives, and most possessive markers, while these suffixes are t-initial with feminine nouns
(3b).
(3)
a. geet-ki
geet-kaŋ
geet-kas
geet-kew
geet-key
geet-ka
geet-’ye
‘the tree’
b. bilaan-ti
‘this tree’
bilaan-taŋ
‘that tree’
bilaan-tas
‘which tree’
bilaan-tew
‘my tree’
bilaan-tey
‘your tree’
bilaan-ta
‘his/her tree’
bilaan-tis
bilaan-tie
geet-kaynu ‘our tree’
bilaan-tayno
geet-kiŋ
‘your pl. tree’
bilaan-tiŋ
geet-’yo
‘their tree’
bilaan-tio
‘the woman’
‘this woman’
‘that woman’
‘which woman’
‘my woman’
‘your woman’
‘his woman’
‘her woman’
‘our woman’
‘your pl. woman’
‘their woman’
For the purposes of this paper, I will focus on this type of gender agreement.
As has been documented elsewhere, gender in EOT languages is sensitive to plurality.
This is broadly referred to as “gender polarity”, but it sometimes manifests as neutralization rather than polarity per se. This varies from language to language; below I summarize
the situation in a number of different EOT languages.
In Standard Somali (SS), according to Saeed (1999: 54–55), most plural nouns reverse
their gender. There are multiple different plural suffixes, and several of them trigger
gender polarity, as can be seen in (4a). However, plurals of masculine nouns formed by
reduplication, as well as ‘a subgroup of masculine suffixing nouns’ retain their masculine
gender in the plural, as seen in (4b).
207
Mary Paster
(4)
a. abtí (m)
túke (m)
káb (f)
galáb (f)
b. wán (m)
béer (m)
dhéri (m)
wáran (m)
‘maternal uncle’
‘crow’
‘shoe’
‘afternoon’
‘ram’
‘liver’
‘clay pot’
‘spear’
abti-yó (f)
tuka-yáal (f)
kab-ó (m)
galb-ó (m)
wan-án (m)
beer-ár (m)
dhery-ó (m)
warm-ó (m)
‘maternal uncles’
‘crows’
‘shoes’
‘afternoons’
‘rams’
‘livers’
‘clay pots’
‘spears’
Thus, in SS, in the plural all feminine nouns become masculine, and some masculine
nouns become feminine but some stay masculine.
Central Somali exhibits a different pattern, where plurals formed with the suffix -o
exhibit polarity (5a), while plurals formed with -(i)yaal are masculine regardless of their
gender in the singular (5b) (Saeed 1982: 11–12).
(5)
a. fileer-taas (f)
laan-taas (f)
shiid-kaas (m)
eleeŋ-kaas (m)
b. jeer-taas (f)
shimbir-taas (f)
ba’iid-kaas (m)
weer-kaas (m)
‘that arrow’
‘that branch’
‘that stone’
‘that ram’
‘that hippo’
‘that bird’
‘that oryx’ ba’
‘that jackal’
fileer-o-gaas (m)
laam-o-gaas (m)
shiid-o-daas (f)
eleem-o-daas (f)
jeer-iyaal-kaas (m)
shimbir-iyaal-kaas (m)
iid-iyaal-kaas (m)
weer-iyaal-kaas (m)
‘those arrows’
‘those branches’
‘those stones’
‘those rams’
‘those hippos’
‘those birds’
‘those oryxes’
‘those jackals’
Lecarme’s (2002) discussion of an unidentified Somali dialect (which appears to be
distinct from both SS and CS) includes the observation that each of several different pluralization strategies tends to result in plural forms with a particular gender. For example,
all plurals in -o (whose singulars are mostly feminine, but some masculine nouns also
occur in this group) (6a) are masculine (Lecarme 2002: 118). Plurals in -oyin (whose singulars are always feminine) are always masculine (6b) (Lecarme 2002: 119). Plurals in -yaal
(6c) (whose singulars can be masculine or feminine) are “masculine or feminine, depending on regional variation, and thus either polaric or not” (Lecarme 2002: 119) (though
note that the only plural forms provided are feminine).
(6) a. fár (-ta) (f)
náag (-ta) (f)
maálin (-ka) (m)
wáran (-ka) (m)
208
‘finger’
‘women’
‘day’
‘spear’
far-ó (-á-ha) (m)
naag-ó (-á-ha) (m)
maalm-ó (-á-ha) (m)
warm-ó (-á-ha) (m)
‘fingers’
‘women’
‘days’
‘spears’
13 Gender instability in Maay
b. hóoyo (-áda) (f)
eeddó (-áda) (f)
magaaló (-áda) (f)
xeró (-áda) (f)
c. maroodí (-ga) (m)
waraábe (-áha) (m)
áabbe (-áha) (m)
jáalé (-áha) (m)
‘mother’
‘paternal aunt’
‘town’
‘enclosure’
‘elephant’
‘hyena’
‘father’
‘comrade’
hooyo-óyin (-ka) (m)
eeddo-óyin (-ka) (m)
magaalo-óyin (-ka) (m)
xero-óyin (-ka) (m)
maroodi-yáal (-sha)
waraaba-yáal (-sha) (f)
aaba-yáal (-sha) (f)
jaala-yáal (-sha) (f)
‘mothers’
‘paternal aunts’
‘towns’
‘enclosures’
‘elephants’
‘hyenas’
‘fathers’
‘comrades’
Lecarme’s analysis locates gender features in the various plural suffixes, explaining
the connection between the gender changes and the use of each of the suffixes.
In Maay, plural nouns are masculine, regardless of whether the plural is formed with
the suffix -o (7a), the suffix -yal (7b), or with both (7c) (Paster 2007; Comfort & Paster
2009; Paster 2010) (note that /k/ and /t/ lenite to [ɣ] and [ð] intervocalically).
(7)
a. d͡ʒeer-tey (f)
gewer-tey (f)
walaal-key (m)
ad͡ʒir-key (m)
b. mindi-ðey (f)
gaʔan-tey (f)
bakeeri-ɣey (m)
miis-key (m)
c. d͡ʒeer-tey (f)
gaʔan-tey (f)
ad͡ʒir-key (m)
miis-key (m)
‘my hippo’
‘my daughter’
‘my brother’
‘my thigh’
‘my knife’
‘my hand’
‘my cup’
‘my table’
‘my hippo’
‘my hand’
‘my thigh’
‘my table’
d͡ʒeer-o-ɣey (m)
gewer-o-ɣey (m)
walaal-o-ɣey (m)
ad͡ʒir-o-ɣey (m)
mindi-yal-key (m)
gaʔan-yal-key (m)
bakeeri-yal-key (m)
miis-yal-key (m)
d͡ʒeer-o-yal-key (m)
gaʔam-o-yal-key (m)
ad͡ʒir-o-yal-key (m)
miis-o-yal-key (m)
‘my hippos’
‘my daughters’
‘my brothers/sisters’
‘my thighs’
‘my knives’
‘my hands’
‘my cups’
‘my tables’
‘my hippos’
‘my hands’
‘my thighs’
‘my tables’
In Tunni, the “gender-opposition is neutralized in Plural nouns, all of which are Masculine” (Tosco 1997: 43; no illustrative examples are provided). And finally, Rendille exhibits
a complex pattern where some nouns appear to switch gender in the plural, while others
take a separate set of plural-agreeing suffixes beginning with /h/ rather than feminine
/t/ or masculine /k/ (see Oomen 1981 for much more detailed discussion).
Summing up, it can be observed that some EOT languages have true polarity and others have only masculine plurals. None of the languages (as far as I am aware) have only
feminine plurals.1 It seems likely that Proto-EOT did exhibit polarity, since polarity ex1A
reviewer points out that K’abeena has feminine plurals. According to Mous (2008: 143), plurals “trigger masculine agreement in the demonstratives, [plural] agreement in the definite markers, but feminine
agreement on the verb in external, clausal agreement.” Since I am focusing on agreement with determiners,
demonstratives, and possessors, for the present purposes I would not consider K’abeena to have feminine
plurals.
209
Mary Paster
ists in Cushitic languages outside of the EOT group as well, e.g., in Oromo (Andrzejewski
1960) and Burunge (Wolff 2014).
This background on gender phenomena in EOT will be relevant to the discussion of
gender instability in §4, since I will argue that it is the masculine-plural pattern in Maay
that provides the mechanism for the emergence of instability.
4 Gender instability in Maay
Wolff (2014) observes that “[a] notable historical feature [of Afro-Asiatic languages] is
‘gender stability’, meaning that words for common things tend to share the same gender
across the languages of the Afro-Asiatic phylum, no matter whether or not the particular
words are cognate across the specific languages in question” (the implication being that
Afro-Asiatic exhibits a greater degree of gender stability than other families do). But as I
will show below, there are exceptions to the ‘gender stability’ generalization within the
Maay language itself.
Considering a sample of 55 common lexical items in Maay, each elicited from up to 6
speakers, I found 34 of them to be consistently masculine, shown in (8) (note that animals
are deliberately excluded, for reasons to be explained below).
(8)
Consistently masculine nouns2
moos-ki
doŋ-ki
baaka-ɣi
hɛɛl-ki
belet-ki
hawuug-i
ilbap-ki
ɛɛs-ki
miniŋ-ki
nal-ki
af-ki
los-ki
biiŋ-ki
wo𝛽i-ɣi
haðag-i
miis-ki
gɛɛt-ki
3
‘the banana’
‘the boat’
‘the box’
‘the cardamom’
‘the city’
‘the corn’
‘the door’
‘the grass’
‘the house’
‘the light’
‘the mouth’
‘the peanut’
‘the pin’
‘the river’
‘the rope’
‘the table’
‘the tree’
suŋ-ki
buug-i
kawaʃ-ki
kuraas-ki
nard͡ʒiŋ-ki
bakeri-ɣi
dɛp-ki
maða-ɣi
fur-ki
beer-ki
basal-ki
galaŋ-ki
barit-ki
d͡ʒit-ki
kasab-ki
nyaanya-ɣi
hidig-i
‘the belt’
‘the book’3
‘the cabbage’
‘the chair’
‘the coconut’
‘the cup’
‘the fire’
‘the head’
‘the key’
‘the liver’
‘the onion’
‘the pen’
‘the rice’
‘the road’
‘the sugarcane’
‘the tomato’
‘the star’
Masculine nouns ending in k appear to have the suffix -i rather than -ki because the /k/ of the stem and
/k/ of the suffix reduce to a single /k/ and then optionally undergo intervocalic voicing. This stop does not
undergo intervocalic lenition to [ɣ] because lenition applies before degemination. Paster (2007) provides a
deeper discussion of Maay phonology.
210
13 Gender instability in Maay
Six nouns in the dataset were consistently feminine, shown in (9).
(9) Consistently feminine nouns
saɁad-i
far-ti
luk-ti
‘the clock’4
‘the finger’
‘the leg’
ɗɛk-ti
saan-ti
suk-ti
‘the ear’
‘the footprint’
‘the market’
Interestingly, 15 nouns in the dataset showed inconsistent gender across speakers, as
shown in (10).
(10) Unstable nouns
ukun-ti ~ ukuŋ-ki
il-i ~ il-ki
beer-ti ~ beer-ki
gaɁan-ti ~ gaɁan-ki
mindi-ði ~ mindi-ɣi
embe-ði ~ embe-ɣi
istaraʃa-ði ~ istaraʃa-ɣi
irbid-i ~ irbit-ki
‘the egg’
‘the eye’5
‘the garden’
‘the hand’
‘the knife’
‘the mango’
‘the napkin’
‘the needle’
farketi-ði ~ farketi-ɣi ‘the fork’
sun-ti ~ suŋ-ki
‘the poison’
dariʃa-ði ~ dariʃa-ɣi ‘the window’
siin-ti ~ siiŋ-ki
‘the hip’
hambal-i ~ hambal-ki ‘the leaf’
kaal-i ~ kaal-ki
‘the spoon’
baloon-ti ~ baloon-ki ‘the ball’
In the following section, I will propose an analysis of the unstable nouns in (10), attempting to explain how their gender came to be ambiguous across speakers. I will suggest
that the pattern of gender neutralization to masculine in Maay plurals creates ambiguity in the gender of singular nouns (particularly when the plural form is more familiar),
leading speakers to sporadically reassign the gender of some nouns.
5 Analysis
To explain the unstable nouns, we could look at properties of both the speakers and the
nouns themselves. Before presenting my proposed analysis that attributes the instability
to gender neutralization in the plural, I consider a number of other factors that could
potentially be relevant to gender instability, showing that none turns out to provide an
explanation for the observed phenomenon.
To start, we might ask whether age, gender, region, or language use could explain
the divide among the speakers. Table 1 provides some demographic and language use
information for each of the Maay speakers consulted for this project.
4
Feminine nouns ending in t appear to have the suffix -i due to degemination as with masculine nouns
ending in k.
5 Feminine nouns ending in l appear to have the suffix -i due to a regular phonological rule that deletes /t/
after /l/.
211
Mary Paster
Table 1: Demographic and language use information
Speaker
Gender
Age
Origin
OM
M
33
Kowan (near Jamaame)
HJ
M
30s
Jamma; lived in Kenya
JA
M
62
Jilib
HM
M
42
Jamaame; grew up in
Mogadishu
BM
F
48
Jamaame
LJ
M
52
Jamaame
AM
M
27
Kismaayo; lived in Kenya
HA
F
30s
MA
F
23
Jamaame; grew up in
Kenya
Jamaame
KJ
F
50s
Jamaame; Kismaayo; lived
in Kenya
Languages
Zigua, Swahili, English
Zigua, English, Somali
Some English, some
Somali
Somali, Zigua, English,
Swahili
Zigua, Somali, English,
some Swahili
Zigua, Somali, English,
Swahili, Italian, some
Spanish
Zigua, Somali, English,
Swahili, Turkana,
Giryama
Zigua, Swahili, English
Zigua, Swahili, Somali,
some English
Zigua, Somali, Swahili,
some English
In Table 2, I give the gender of each noun according to the available data elicited from
each speaker.
Considering the noun gender data by speaker, a few generalizations emerge. First, the
available data from LJ indicate that he has assigned masculine gender to all of the nouns
in question. Second, HA also has many of these nouns as masculine, and HJ has some
masculine. All available instances of these 15 words from AM and OM were feminine.
In an attempt to align these observations with the speakers’ demographic characteristics, it can be observed that LJ, HA, and HJ, who had more masculine forms of these
nouns than other speakers did, are from the same area. However, OM and BM are also
from the same region and had almost all of these nouns as feminine, so an explanation in
terms of a geographically defined dialect feature is unlikely. The age of the speakers also
does not seem to be a relevant factor, since the group of three speakers who produced the
212
13 Gender instability in Maay
Table 2: Noun gender by speaker
LJ
ball
egg
eye
fork
garden
hand
hip
knife
leaf
mango
napkin
needle
poison
spoon
window
HJ
M
F
M
M
M
M
F
M
M
F
F
M
M
HA
AM
OM
M
M
M
M
M
F
M
M
F
M
M
M
M
M
F
F
F
F
BM
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
M
F
most masculine forms includes one of the oldest speakers (LJ) and two of the youngest
(HA and HJ). Gender seems irrelevant, since LJ and HJ are male while HA is female. And
finally it can be observed that the two speakers who produced all feminine forms, OM
and AM, are younger men, but HJ is also a younger man and produced several masculine
forms. Thus, properties of the speakers themselves do not seem to provide any insight
into the behavior of the unstable nouns.
It is possible that language experience and/or language attitudes play a role in determining which gender each speaker will assign to a given noun, but here again no clear
explanation emerges. In language attitude surveys, HA and HJ, both of whom produced
more masculine nouns in this set than other speakers, stated that they identify more
with Zigua (a Bantu language spoken by many members of the community) than Maay
as their mother tongue and that they speak Zigua at home. However, it is not clear what
influence Zigua might have on the gender of their Maay nouns apart from general interference with regular Maay usage, since Zigua has noun classes rather than binary
gender. No other details about any of the speakers’ experiences with languages other
than Maay appear to correlate with the data.
Having considered the possibility that the gender of the unstable nouns is a dialect
feature relating to the geographic origin of, or other facts about, the speakers and finding
none, we might consider whether there are properties of the nouns themselves that can
shed light on why they are unstable across speakers.
One obvious potential factor could be the phonological form of the nouns, but a consideration of the nouns in (8-10) does not reveal any good candidates for a phonological
213
Mary Paster
property that might unify any set of nouns. The masculine nouns (8) and unstable nouns
(10) end in either a consonant or a vowel, and they can have anywhere from one to
three syllables (one unstable noun has four syllables). The feminine nouns in (9) all end
in a consonant, and only one has more than one syllable, but there are stable masculine nouns and unstable nouns that also have these properties, so the shape of the six
feminine nouns in (9) does not reflect a phonological natural class.
A second property of the nouns that could be considered is whether they are native
or borrowed, and if borrowed, the source of the borrowing. But here again, there is no
clear pattern. Table 3 gives the sources for four of the nouns in the unstable class that
can be identified as borrowings.
Table 3: Sources of borrowed unstable nouns
Noun
Source
ball
fork
napkin
poison
Italian or English
Italian
Italian
Arabic
While it is true that two or three of the unstable nouns are Italian borrowings, one of
them is from Arabic, and the remaining 11 unstable nouns are apparently native. Therefore once again no solid generalization can be made. The list stable masculine nouns in
(8) also includes both native words and borrowings from English, Italian, and Arabic; the
stable feminine class in (9) includes both native words and Arabic borrowings.
A third possibility is that the semantics of the unstable nouns might explain their behavior. At first this does not seem to be a likely source for an explanation, since nouns of
many different semantic categories have unstable gender, including body parts, utensils,
foods, and miscellaneous others. Recall that animals were deliberately omitted from this
study. The reason is that if an animal noun exhibits gender variability, this could be attributed to a functional use of gender corresponding to the animal’s sex. In fact, several
animal nouns do show gender variability, as seen in (11).
(11)
yahas-ti ~ yahas-ki
mayoonda-ði ~ mayoonda-ɣi
d͡ʒeer-ti ~ d͡ʒeer-ki
‘the crocodile’
‘the (monkey sp.)’
‘the hippo’
When these forms were produced, it is possible that the speaker had an animal of a
particular sex in mind, even if this was not necessarily indicated in the English translation, (e.g., even if the speaker did not specify ‘a male crocodile’ when giving the form
yahas-ki).
Given this, the set of unstable nouns does not initially seem to form a semantic natural
class. However, a closer look reveals a possible unifying property for several of the unstable nouns. It is instructive to compare the unstable nouns in Maay with the so-called
214
13 Gender instability in Maay
‘double gender’ nouns in Dutch, where Semplicini argues that “nouns whose referents
are characterized by a high degree of individuation tend to trigger common agreement,
while nouns with less individuated referents are more likely to trigger neuter agreement”
(2012: 176). The notion of the “degree of individuation” may be relevant in Maay as well,
since most of the unstable nouns in (10) are at least somewhat likely to occur frequently
in the plural, and furthermore, some (especially paired body parts) probably occur much
more often in the plural form than the singular. This observation about the unstable
nouns is noteworthy because it links gender instability to the masculine-plural pattern
in Maay: because the masculine-plural pattern makes the gender of singular nouns unrecoverable from their plurals, nouns that are frequently plural may be susceptible to
gender instability.
Supposing that the gender of a noun’s singular form is determined by analogy using
the plural when the gender of the singular noun is unknown, speakers will arrive at
different singular forms depending on which words they choose to form the analogy.
For example, when a speaker tries to determine the singular form of ‘hip’ from the plural
form siim-o-ɣi ‘the hips’, if he/she analogizes to sum-o-ɣi ‘the belts’ as in (12a), then the
singular ‘hip’ will end up masculine. However, if the speaker instead analogizes to saamo-ɣi ‘the footprints’ as in (12b), then the singular ‘hip’ will end up feminine.
(12) Two possible analogies for recovering the gender of a noun from its
plural
a. sum-o-ɣi
siim-o-ɣi
b. saam-o-ɣi
siim-o-ɣi
‘the belts’
‘the hips’
‘the footprints’
‘the hips’
(m.) : suŋ-ki
(m.) : siiŋ-ki
(m.) : saan-ti
(m.) : siin-ti
‘the belt’ (m.) ::
‘the hip’ (m.)
‘the footprint’ (f.) ::
‘the hip’ (f.)
The analogy in (12a) may overapply relative to (12b) due to a preponderance of masculine nouns in the lexicon, since if a speaker selects a phonological neighbor to form
the analogy, more often than not the neighbor will happen to be masculine. This would
cause feminine nouns to shift to masculine more often than the reverse, although shifts
could occur in either direction.6
6 Conclusion
A number of predictions follow from the hypothesis presented above. Four predictions
are enumerated in (13); I will discuss each in more detail below, showing that the available
evidence is consistent with the predictions.
6
Note that my analysis does not hinge on the proposal that analogy is to phonological neighbors in particular.
The analogy could instead be to semantic relatives, for example, and the same result would be predicted to
emerge.
215
Mary Paster
(13)
a. More unstable nouns should have feminine cognates in related languages
than masculine cognates
b. More speakers will have the unstable nouns as feminine than masculine
c. Nouns that are frequently plural are most likely to change gender in
languages with the masculine-plural pattern
d. Languages with the masculine-plural pattern are more likely to have gender
instability
The first prediction, that unstable nouns should have feminine cognates in related languages, follows from the observation made earlier that, given the scenario I have outlined
for how singular forms are recovered from their gender-ambigious plural counterparts,
the overall masculine skewing of the lexicon will cause more feminine nouns to shift to
masculine than the reverse. Thus, the instability will more often affect historically feminine nouns. A look at cognates in EOT does appear to uphold this prediction, although
the sample size is small. For example, in Central Somali, all of the cognates of the unstable nouns given by Saeed (1987) are feminine (‘egg’, ‘eye’, ‘hand’, and ‘leaf’). None of the
other nouns in the unstable class have Central Somali cognates provided (‘knife’ is also
feminine in Central Somali, but it is not a cognate with the Maay form). Similarly, in
Tunni (Tosco 1997), ‘egg’, ‘eye’, ‘hand’, ‘knife’, ‘leaf’ are all feminine (no other cognates
are provided). Given the skewing of stable Maay nouns towards masculine, it is striking
that all of the cognates that were found in Central Somali and Tunni for the unstable
nouns are feminine. Thus, the prediction appears to be accurate, though a consideration
of cognates in other EOT languages is warranted.7
The second prediction, that more speakers will have the unstable nouns as feminine
than masculine, also relates to the idea that the analogical recovery of singular gender
will tend to shift previously feminine forms into the masculine category. If this is indeed
the mechanism producing the instability, we expect to see a recurring pattern where
most speakers have a particular noun as feminine but one or more speakers innovates a
masculine form. In that case we expect to see a recurring pattern where most speakers
continue to treat a given noun as feminine, while a smaller number of speakers treat it as
masculine. Again the data do appear to uphold this prediction, though again the sample
is small. For all but two of the unstable nouns (‘mango’ and ‘poison’), at least as many
speakers have the noun as feminine as masculine. And in several cases the feminines
outnumber the masculines by a ratio of at least three to one. ‘Mango’ and ‘poison’, while
they do not uphold the predicted trend themselves, are still not problematic since the proposed mechanism does allow masculine nouns to shift to feminine. A finding that these
7
Note that masculine nouns can also become unstable through the same mechanism; they are just posited
to be less likely to do so than feminine nouns. Note also that a noun that is not more commonly attested
in the plural form than in the singular can still become unstable if the speaker does not know its gender;
in that case, rather than analogizing from a plural form, the speaker might just guess at the noun’s gender.
Thus, although I am suggesting that a feminine noun that is frequently used in the plural and rarely in
the singular is the most likely type of noun to become unstable, other types of nouns may also become
unstable.
216
13 Gender instability in Maay
two nouns have masculine cognates in EOT would add further weight to the conclusion
that this second prediction is upheld in the data.
The third prediction was that nouns that are frequently plural are most likely to change
gender in languages with the masculine-plural pattern. This is a cross-linguistic prediction that could not be tested within the scope of this study. Within the EOT group, Tunni
is the other language that has a uniform masculine-plural pattern, so one might expect
to find a similar pattern of noun gender instability in that language, whereas languages
with true gender polarity (like Rendille) would not be as likely to have gender instability
since the gender of the singular form is unambiguously recoverable (though reversed)
from the plural. I am not aware of the existence of multi-speaker gender-marked noun
datasets for these or other relevant languages that would allow us to test this prediction
cross-linguistically at present, but as I have argued here, it does seem to be true for Maay.
The final prediction was that languages with the masculine-plural pattern are more
likely to have gender instability than other languages are. As with the previous prediction, this is a cross-linguistic prediction that has yet to be tested, though I have argued
that Maay is an example of a language upholding this prediction. It is possible that the
lack of the Maay-type masculine-plural pattern elsewhere in Cushitic enables us to reconcile the gender instability in Maay with Wolff’s assertion, cited at the beginning of
this paper, that Afro-Asiatic in general exhibits gender stability, but this remains an
empirical question to be tested by comparing Maay and other languages that have the
masculine-plural pattern with those that do not.
As a whole, then, the explanation I have proposed for the gender instability in Maay
nouns does find support within the language and tentatively within the EOT group. Its
applicability outside of EOT and Cushitic in general remains to be tested.
A final observation is that a number of languages have genderless plurals, and this
analysis of Maay does predict that such languages should be susceptible to gender instability.8 It is possible that further research will reveal that gender polarity has indeed
developed in such languages; it is also conceivable that there are additional factors that
have facilitated its development in Maay that are not present in other languages. For
example, because the speakers are refugees and live in a community where not everyone speaks Maay and where several other languages are used, the language as a whole
could be considered somewhat unstable (which is also consistent with the high degree of
inter-speaker variability discussed in Paster 2013). Perhaps the gender neutralization in
the plural has combined with a generally unstable language situation to produce the phenomenon we observe in Maay. Further research is needed to determine how widespread
the phenomenon is and what determines when and where it emerges.
Acknowledgements
Many thanks to the Somali Bantu Community of San Diego, to my fall 2012 Field Methods
class at Pomona College, and to Rodrigo Ranero and Rebekah Cramerus for participating
8
Thanks to an anonymous reviewer for raising this issue.
217
Mary Paster
in this research. Thanks also to the audience at ACAL 45 and to the anonymous reviewers
of this paper for helpful comments. This research was funded in part by an Arnold L. and
Lois S. Graves Award and by a Wig Teaching Innovation Grant from Pomona College.
References
Andrzejewski, Bogumil. 1960. The categories of number in noun forms in the Borana
dialect of Galla. Africa 30. 62–75.
Comfort, Jade & Mary Paster. 2009. Notes on Lower Jubba Maay. In Masangu Matondo,
Fiona McLaughlin & Eric Potsdam (eds.), Selected Proceedings of the 38th Annual Conference on African Linguistics, 204–216. Somerville, Massachusetts: Cascadilla Proceedings Project.
Corbett, Greville G. & Richard J. Hayward. 1987. Gender and number in Bayso. Lingua
73. 1–28.
Lecarme, Jacqueline. 2002. Gender ‘polarity’: Theoretical aspects of Somali nominal morphology. In Paul Boucher & Marc Plénat (eds.), Many morphologies, 109–141. Somerville,
Massachusetts: Cascadilla Press.
Mous, Maarten. 2008. Number as an exponent of gender in Cushitic. In Erin Shay & Zygmunt Frajzyngier (eds.), Interaction of morphology and syntax: Case studies in Afroasiatic, 137–160. Amsterdam: Benjamins.
Oomen, Antoinette. 1981. Gender and plurality in Rendille. Afroasiatic Linguistics 8(1).
35–75.
Paster, Mary. 2007. Aspects of Maay phonology and morphology. Studies in African Linguistics 35(1). 73–120.
Paster, Mary. 2010. Optional multiple plural marking in Maay. In Franz Rainer, Wolfgang U. Dressler, Dieter Kastovsky & Hans Christian Luschützky (eds.), Variation and
change in morphology (Current Issues in Linguistic Theory 310), 177–192. Amsterdam:
Benjamins.
Paster, Mary. 2013. An I-Language approach to inter-speaker variation in Maay. Lecture
presented at the 44th Annual Conference on African Linguistics/Georgetown University Round Table on Languages and Linguistics, Georgetown University.
Saeed, John I. 1982. Central Somali: A grammatical outline. Monographic Journals of the
Near East: Afroasiatic Linguistics 8(2). 77–119.
Saeed, John I. 1987. Somali reference grammar. Wheaton, Maryland: Dunwoody Press.
Saeed, John I. 1999. Somali. Amsterdam: John Benjamins Publishing Company.
Tosco, Mauro. 1997. Af Tunni: Grammar, Texts, and Glossary of a Southern Somali Dialect.
Köln: Rüdiger Köppe Verlag.
Wolff, H. Ekkehard. 2014. Afro-Asiatic languages. Encyclopedia Brittanica, online edition. http://www.britannica.com/EBchecked/topic/8488/Afro- Asiatic- languages/,
accessed 2014-04-03.
218
Chapter 14
Egyptian Arabic broken plurals in DATR
Lindley Winchester
Georgetown University
This paper examines plural inflectional processes in Egyptian Arabic, with specific focus on
the complex broken plural system. The data used in this examination is a set of 114 lexemes
from a dictionary of the Egyptian Arabic variety by Badawi & Hinds (1986) collected through
comparison of singular to plural template correspondences proposed by Gadalla (2000). The
theoretical side of this analysis builds upon Alain Kihm’s realizational “Root-and-Site Hypothesis”, which categorizes concatenative and non-concatenative morphological processes
as approachable in the same manner when discussing inflection as not only represented in
segments but also as “sites” where inflectional operations may take place (Kihm 2006: 69). To
organize the data through a computational lens, I emulate Kihm’s approach in DATR, a lexical knowledge representation language, to generate the grammatical forms for a set of both
broken and regular plural nouns. The hierarchically-structured inheritance of DATR allows
for default templates to be defined and overridden, permitting a wide scope of variation to
be represented with little code content. Overall, the analysis reveals that complex morphological phenomena, such as the broken plural, can be accounted for through a combination
of theoretical and computational approaches.
1 Introduction
Egyptian Arabic is a branch of the Arabic language and the national language of Egypt.
Outside of Egypt, it is intelligible in other Arabic-speaking countries, such as Libya, Syria,
and Yemen. It is defined characteristically as part of the central and south branch of the
Afroasiatic language family and Semitic genus (Lewis 2009). The particular inflectional
process in focus here occurs on the nominal forms of the language, which are inflected
for plural number through one of two separate processes, a suffixal inflection and infixational inflection, both of which will be elaborated upon further in §2. The latter process
will take the majority of the focus, analyzed through a pre-existing theoretical framework and formalized in the computational model, datr. The purpose of this examination
is to computationally model theory in the construction of broken plurals in Egyptian Arabic, seeking an analysis that encompasses a majority, if not all, of the complex forms in
question.
Lindley Winchester. Egyptian Arabic broken plurals in DATR. in Jason Kandybowicz, Travis
Major, Harold Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected
papers from the 45th Annual Conference on African Linguistics, 219–231. Berlin: Language Science
Press. DOI:10.5281/zenodo.1251734
Lindley Winchester
2 Nominal inflection
Although nominals are inflected for definiteness, possession, number, and grammatical
gender in Egyptian Arabic, the discussion here focuses on the language’s singular and
plural number inflection (Gadalla 2000: 129–130)1 . While singular number is not overtly
marked, the expression of plural number in the Arabic varieties is realizable through
two different inflectional processes and therefore partitions the lexicon of the language
into two groups according to which process they utilize. The group of words which employ the first process, called the sound plurals, add a suffix to the singular stem without
changing its internal structure. This group is loosely analogous to the dog/dog-s number inflection in English. However, unlike English the suffixes which attach to the stem
agree in gender.
The broken plural group (bps from here) is characterized by internal stem modification
through the infixation of interweaving vowels, which vary in both vowel quality, length,
and position between the consonantal roots of the stem. These plurals are considerably
less predictable than their suffixal counterparts, analogous to the irregular man/men inflections in English. An example of this group is the masculine singular noun ʃaahid
‘witness’, which does not attach the masculine suffix /-iin/ but becomes ʃuhuud in the
plural. Unlike verbal derivation, the broken plural inflection cannot be associated with
any one sequence of vowels (such as the -u-uu- format in ʃuhuud) and similarly can
not be defined through the process of allomorphy. Rather, the vowel qualities of both
the singular and plural forms are semi-regular at best, making it difficult to distinguish
any one vowel as the plural marker and any one vowel as the singular (Kihm 2006: 70).
Examples of bp inflectional variation can be seen in Table 1 below.2
Table 1: Examples of bp variation.
Singular
Broken Plural
Gloss
suura
taman
ʃagaan
suwar
ʔatmaan
ʃuguun
‘chapter of the Koran (331)’
‘price (137)’
‘sorrow (453)’
In addition, plural patterns cannot be uniquely associated with any one singular form
nor any singular with any one plural form. For example, the C1 aC2 C3 singular templatic
form is associated with multiple bp patterns as seen in Table 2 below.
1
Egyptian Arabic, unlike Classical and Modern Standard Arabic, does not inflect for case through affixation
(Gadalla 2000: 108).
2
The numbers listed in parentheses after each gloss in this and the following tables refer to the page in
Badawi & Hinds (1986) on which the respective example is found.
220
14 Egyptian Arabic broken plurals in DATR
Table 2: Examples of inflectional variation between templates.
Singular
Plural
Gloss
garh
(C1 aC2 C3 )
guruuħ, giraaħ
(C1 uC2 uuC3 , C1 iC2 aaC3 )
‘wound (153)’
raxw
(C1 aC2 C3 )
raxaawi
(C1 aC2 aaC3 i)
‘whiplash (331)’
2.1 Broken plurals in theory
Currently in Arabic linguistics, morphological research has been divided into two camps
by differing theoretical approaches. Previously, the field assumed a root-based approach
used by traditional Arab grammarians in explaining Arabic morphology. In opposition
to this traditional approach are the word or stem-based approaches (Ratcliffe 2013: 71–
91). From the span of approaches used to analyze bps,3 I selected Kihm’s (2006) analysis
of bps and verbal nouns within Classical Arabic to provide the main theoretical framework in this paper.4 This decision was influenced heavily by Kihm’s adherence to the
traditional root-and-pattern approach to Arabic morphological studies in addition to its
flexibility and adaptability to datr.
The widely accepted approach on the opposing side, a prosodic approach by McCarthy
& Prince (1990), would not satisfy the intended goal of this paper. In their analysis, the
main focus is placed on the leftmost heavy syllable, or two moras, as the singular stem’s
minimal word within which the bp is formed (Ratcliffe 1998: 80; McCarthy & Prince 1990:
231). With this, they structure their analysis around developing bps from lexemes’ singular stems, replacing some material while utilizing portions of its structure as distinctive
in developing the iambic plural structure. One such feature that is transferred from the
singular to plural form is said to be the vowel length of the final syllable when the singular’s first syllable is heavy (CVC or CVV). However, despite being the “most familiar of
the non-root properties,” it is not consistently maintained in EA data, as seen in the singular ʃaahid becoming plural ʃuhuud ‘witnesses’ (McCarthy & Prince 1990: 218; Badawi
& Hinds 1986: 122). Though ʃaahid does contain an initial heavy syllable /ʃaa-/, the short
vowel in the singular’s final syllable /-hid/ is not maintained in the plural but rather is
lengthened to a long vowel.5 Though rules such as this do find some grounding in the EA
data at hand, they are not consistent enough to develop wide sweeping generalizations.
Furthermore, McCarthy & Prince’s analysis places a heavy emphasis on the iambic
3
Namely, Hammond (1988), McCarthy & Prince (1990), Ratcliffe (1998), and Kihm (2006). This list is by no
means exhaustive.
4
With supplemental information from an array of his other publications, each of which assists in explaining
the framework further.
5
This is just one feature McCarthy & Prince (1990) discuss as transferrable from the singular stem to a bp.
Refer to Kihm (2006) for a further elaboration on the issues with relying on singular forms in determining
bps.
221
Lindley Winchester
plural, which they present as most productive in the Arabic lexicon. Though this may be
true, the analysis presented is not easily adaptable to the remaining non-iambic forms
in my data and therefore cannot serve for the purpose of this paper. The root-based
assumptions used in Kihm’s analysis allow for more flexibility in presenting a wider
array of the type of bps found in the data.
Finally, Kihm’s theoretical adherence to the root-and-pattern approach also allows for
an easy transition into datr, which focuses on the lexeme, defined as the consonantal
root for this paper, rather than the morpheme as the minimal sign in a morphological
paradigm (Brown & Hippisley 2012: 5).
2.2 The Root-and-Site Hypothesis
Kihm’s Root-and-Site Hypothesis (rsh) takes a realizational nonsegmental concatenative approach to the bp phenomenon in Classical Arabic based upon a hpsg-type lexicon
(see Pollard & Sag 1994). He argues this and other non-concatenative morphological processes could be absorbed into the category of concatenative morphology, shared by the
sound plural inflection, if not only segments but also abstract elements, which he names
functional “sites,” can act as the locations in which morphology can occur (Kihm 2006:
69). These locations can be both outside and inside the stem boundary.
The functional site designated for the nominal bp inflection is located within the stem,
between the second and third consonants. Though root-based, this placement coincides
well with the word/stem-based hypotheses from Ratcliffe (1998) and McCarthy & Prince
(1990). This root internal site is thus associated with the feature bundle num(ber) and is
realized by the insertion of a glide, designated as /I/ (which can surface as /i/ or /j/), /U/
(which can surface as /u/ or /w/) and /A/ (which can surface as /a/ or /ʔ/) (Kihm 2006:
80).
Once inserted, the featured glide can either remain or spread into a short or long vowel
construction within the word form (Kihm 2006: 80). The determination of which form
surfaces is dependent upon the type of location it is inserted into: a slot designated for
consonants or vowels. It surfaces as a long vowel when inserted in a consonantal location,
and a short vowel when in a vowel slot (Kihm 2006: 81). This short vowel occurrence
accounts for the construction of non-iambic broken plurals (see McCarthy & Prince 1990)
and forms the basis for the “No long vowel inflection” class in the organization of data
for this research.
Defining the diversity of the glide’s timbre and the location in which it is inserted
(whether a consonant or vowel slot) as irreducible, Kihm posits that each lexical entry
must therefore supply the timbre of the glide, the type of slot in which it will be inserted,
and the consonantal roots (Kihm 2006: 81).
3 Methodology
The data collected for the purpose of this research is a summation of a comparative analysis between two written sources. Gadalla’s (2000) comparative morphological analysis
222
14 Egyptian Arabic broken plurals in DATR
of Modern Standard Arabic and Egyptian Arabic supplies a complete list of singular to
broken plural templates (as well as those apt to take the sound plural) for Egyptian Arabic, such as follows: C1 aC2 C3 → C1 uC2 uuC3 , C1 iC2 aaC3 , etc.6 In order to collect a set
of concrete wordforms for analysis, I matched the list of template correspondences to
vocabulary entries listed in Badawi & Hinds (1986) A Dictionary of Egyptian Arabic, in a
similar fashion to:
(1)
C1 aC2 C3 → garħ ; C1 uC2 uuC3 → guruuħ
The collection process resulted in 114 individual lexemes that form bps. These sets are
meant to exhibit the range of variation seen in the broken plural formation from singular
stems in Egyptian Arabic and are not based upon type or token statistical frequency
within the language. The lack of such statistics should be considered a limitation at this
point as the data does not provide a picture of the more or less commonly used bp forms
within the language. However, the purpose of this analysis is not to discuss the most
frequent forms in comparison to their infrequent siblings but rather to encompass as
much of the found variation as possible within the computational construction.
Coinciding with Kihm’s theoretical approach, I have categorized the data into inflectional classes based on their inflection site (at this point assumed to be a long vowel)
in the bp form. These classes are then further separated based on major alterations to
the stem during the inflection process, such as the insertion of a glottal stop prefix or
a non-root based glide. From the 114 sets of singular to plural forms collected, one representative set is selected for each inflection class and subclass, characterized by the
placement of the bp inflection site (class), any modification to the stem (subclass), and
number of consonantal roots. These categories are displayed in Table 3 below, containing examples from Badawi & Hinds (1986), which I have organized according to the site
in which their inflection occurs.
4 datr
datr is a lexical knowledge representation language used to express default-inheritance
networks. Its primary use is the “representation of lexical entries for natural language
processing” (Evans & Gazdar 1989: 66). Therefore, in datr’s language, I am able to define
connections between a lexical entry’s informational content and various nodes, which
contain separate collections of internally related grammatical information, to construct
a representation of the singular and bp forms. My representation heavily relies on networks of inheritance and the specification of morphosyntactic features through attribute
paths. To elaborate, attribute paths can be realized as values, as in an atom: <path1> ==
value, a separate path: <path2> == <path1>, or as a combination of the two: <path3>
== <path1> a. This final example might represent the fact that some morphosyntactic
6 In
alignment with other researchers, Gadalla utilizes F-ʕ-L as markers of the consonantal roots in the
Arabic languages, correlational to C1 -C2 -C3 . For the remainder of this paper, I will use the latter form of
consonantal notation.
223
Lindley Winchester
Table 3: Nouns covered by second datr theory.
Designated Inflection
class
Triconsonantal Roots
Inflected after C1
Inflected after C2
→ with glottal stop prefix
→ with glide insertion
→ previously defined
“derived noun”
Inflected after C3
No long vowel inflection
Quadriconsonantal
Roots
Inflected after C2
No long vowel inflection
Singular form
Plural form
Gloss
sajjid
saada
gabal
garħ
ʃagaan
taman
gibaal
guruuħ
ʃuguun
ʔatmaan
‘male polite form of address (440)’
‘hill (148)’
‘wound (153)’
‘sorrow (453)’
‘price (137)’
garħa
matgar
gawaariħ
mataagir
‘carnivore (153)’
‘place of business (122)’
marsa
ɣuraab
sˤadiiq
dibb
maraasi
ɣirbaan
ʔɑsˤdiqɑɑʔ
dibab
‘harbor (337)’
‘crow (619)’
‘friend (499)’
‘bear (275)’
tuzluk
zooraq
sˤɑjdɑli
tazaalik
zawaariq
sˤɑjɑdlɑ
‘leather leggings (128)’
‘small boat (386)’
‘pharmacist (516)’
feature, named path3, is realized as whatever form path1 realizes plus a word-final /-a/
suffix (Evans & Gazdar 1996: 167–168). For a concrete example, refer to the basic lexical
entry for the noun, gabal ‘hill/mountain’ (gibaal ‘hills/mountains’ for the plural) below
in (2).
(2)
gabal ‘hill/mountain’ lexical entry
GABAL:
<syn_cat> == \isi{noun}
<gender> == masc
<gloss> == hill , or , mountain
<vowel sg> == V2:<vowel>
<vowel pl> == <vowel sg>
<c 1> == g
<c 2> == b
<c 3> == l
<stem sg> == SINGULAR:<stem sg 3>
<stem pl> == INFLC2.
224
14 Egyptian Arabic broken plurals in DATR
Here, I have designated the syntactic category for gabal as a noun, the gender as
masculine, and so forth. The < > denote paths that are realized by the values following
the == (Evans & Gazdar 1996: 169).
Following the conceptual purpose of the datr language, which is to create widesweeping generalizations within language inflection while avoiding redundancy throughout the coding process, we see that the singular stem <stem sg> and plural stem <stem
pl> values refer to a separate node and does not simply state the singular and plural stem
(Evans & Gazdar 1996: 169). It would be contradictory to our purpose as well as datr’s
to simply state <stem sg> == gabal and <stem pl> == gibaal. Rather, as seen in (2), we
rely on networks of inheritance to form these for the lexical entry, and hopefully various
others, by creating separate nodes called singular for the singular stem and inflc2 for
the plural. The paths that are realized by these two nodes can construct the singular and
plural stems through the insertion of vowels and the consonantal root values specified
within the lexical entry. This inheritance appears just as in Figure 1 below, where the
lexical entry gabal looks to the inflc2 node searching for a path that matches its own
<stem pl>. This <stem pl> within inflc2 then yields a template in which gabal inserts
the values for its consonantal roots.
GABAL
INFLC2
stemPL
gVbVVl
Figure 1: gabal inheritance visual representation.
Within the same lexical entry, the singular path is realized by a separate node titled
singular, which itself contains a path designated as <stem sg 3>. Similar to the description above for Figure 1, to form gabal’s singular stem, it looks to a node called singular,
finds a path within it named <stem sg 3>, inherits the template specified there and inserts
its consonantal root values.
The coding within these two nodes can be seen below in (3).
(3)
Singular and post-C2 bp inflection coding
SINGULAR:
<stem sg 3> == ”<c 1>” ”<vowel sg>” ”<c 2>” ”<vowel sg>” ”<c 3>”.
INFLC2:
<stem pl> == ”<c 1>” ”<vowel pl 2>” ”<c 2>” ”<vowel pl>”
”<vowel pl>” ”<c 3>”.
225
Lindley Winchester
The question now is how to associate the appropriate vowel qualities within the <vowel
sg> and <vowel pl> paths defined in the <stem pl> and <stem sg 3> templates. While
it would be easy to simply place them within the templatic structures specified within
the inflc2 and singular nodes, allowing the lexical entry to inherit both the templatic
form and vowel qualities together, the theory would no longer have the ability to account for words that have the same template but different vowels. An example of this is
the singular ʃagaan ‘sorrow’ becoming the plural ʃuguun ‘sorrows’. In the plural, gibaal
and ʃuguun share the same template (C1 -V-C2 -V-V-C3 ) but vary in vowels. In order for
the lexical entries shagaan and gabal to both inherit from the same inflc2 <stem pl>,
the vowel qualities for the respective plural stems must simply be specified in a separate
location where they can be inherited by the corresponding lexical entry. In singular’s
template, we see the vowels are both specified as some default singular vowel (<vowel
sg>), whereas the vowels in inflc2 are designated as a long default plural vowel in the
second syllable (<vowel pl> <vowel pl>) and a non-default plural vowel in the first syllable (<vowel pl 2>). These specifications require that the lexical entry realize the paths:
<vowel sg> and <vowel pl>. By having content from multiple nodes converge into one,
the lexical entry, the result is called a multiple inheritance network (Evans & Gazdar 1996:
202–203). Since a <sg> and <pl> inheritance for the vowels is distinguished, the theory
can link the values from separate vowel nodes to the appropriate singular and plural
vowel paths specified in the templatic structures of the singular and inflc2 nodes.
Referring back to gabal’s lexical entry, we see the singular vowels are to be assigned
from the v2 node and any of its <vowel> path values. Looking at v2, we find the coding
in (4).
(4)
v2 node
V2:
<vowel> == a
<vowel 2> == i.
With this vowel value and datr’s use of multiple inheritance, we can now insert material into the <vowel sg> paths in <stem sg 3>’s template in (3) to create the full singular
stem, g -a-b-a-l → gabal. Since <stem sg 3> does not call for a <vowel sg 2>, the information provided by <vowel 2> in v2 is ignored for now.
For the plural, the vowel values are assigned from the same node as the <vowel sg>’s
path. Using the same procedure as above, datr inserts this <vowel> value into the
<vowel pl> path locations specified in inflc2’s <stem pl> template. Unlike the singular, the template now calls for a <vowel pl 2> value and therefore inserts the values
specified for <vowel 2> within v2, creating g -i-b-a-a-l → gibaal. The full inheritance
hierarchy for gabal can be seen below in Figure 2.
Now let’s expand our theory to account for shagaan. Its lexical entry appears as
below in (5).
226
14 Egyptian Arabic broken plurals in DATR
GABAL
VSG
VPL
SINGULAR
INFLC2
stemSG3
stemPL
gabal
gibaal
Figure 2: gabal inheritance visual representation.
(5) shagaan ‘sorrow’ lexical entry
SHAGAAN:
<syn_cat> == \isi{noun}
<gender> == masc
<gloss> == sorrow
<vowel sg> == V1:<vowel>
<vowel pl> == V6:<vowel>
<c 1> == ʃ
<c 2> == g
<c 3> == n
<stem sg> == SINGULAR:<stem sg 5>
<stem pl> == INFLC2.
As with gabal, the lines designating the syntactic category, gender, gloss, consonantal roots, and stem/vowel qualities for both the singular and plural stems are included.
shagaan follows the same procedure as gabal in forming the plural, inheriting the same
<stem pl> template from inflc2 and plural vowels from a node named v6, which supplies
the /u-uu/ vowel melody. The coding for v6 appears the same as v2, except specifying
<vowel> == u in this instance.
The two lexical entries differ in their singular form and therefore inherit different
templates within the singular node. Specifically, shagaan inherits from a path named
<stem sg 5>, with the number only distinguishing the different templates with no relation to hierarchy. The coding for the singular node now appears in (6), including both
shagaan and gabal’s singular stem formations.
227
Lindley Winchester
(6) singular node coding
SINGULAR:
<stem sg 3> == ”<c 1>” ”<vowel sg>” ”<c 2>” ”<vowel sg>” ”<c 3>”
<stem sg 5> == ”<c 1>” ”<vowel sg>” ”<c 2>” ”<vowel sg 2>”
”<vowel sg 2>” ”<c 3>”.
The remainder of the inheritance remains the same as in gabal. shagaan inherits
singular’s <stem sg 5> and inserts its consonantal roots and the inherited vowel from
the v1 node’s <vowel>, the coding of which is seen below in (7).
(7)
v1 node
V1:
<vowel> == a.
These are then inserted into the appropriate slots in the singular stem’s template (C1 -VC2 -V-V-C3 ), creating ʃ-a-g-a-a-n → ʃagaan. The redundancy we see in comparing nodes
v1 and v2 is necessary in order to capture the vowel variation seen in stems like singular
gabal → plural gibaal, ʃagaan → ʃuguun, and matgar ‘place of business’ → mataagir
‘places of business’. Lexical entries for gabal and matgar will inherit from v2 to achieve
the /a-i/ or /i-a/ vowels in their plural while shagaan inherits from v1 to achieve solely
/a/ vowel insertion. Mimicking gabal above, the tree representation for shagaan can
be seen in Figure 3 below.
SHAGAAN
VSG
VPL
SINGULAR
INFLC2
stemSG5
stemPL
ʃagaan
ʃuguun
Figure 3: shagaan inheritance visual representation.
In addition to triconsonantal roots, the theory can also form quadriconsonantal singular and bp forms. An example of the lexical entry for such can be seen in (8) below.
228
14 Egyptian Arabic broken plurals in DATR
(8)
zooraq ‘small boat’ lexical entry
ZOORAQ:
<syn_cat> == \isi{noun}
<gender> == masc
<gloss> == small, boat
<vowel sg> == V4:<vowel>
<vowel pl> == V2:<vowel>
<c 1> == z
<c 2> == w
<c 3> == r
<c 4> == q
<stem sg> == QUAD_SINGULAR:<stem sg 2>
<stem pl> == QUAD_PL_INFLC2:<stem pl 2>.
It appears the same as the two previous entries but with an additional consonant specified as <c 4>. This lexeme is particularly interesting for containing a glide as its second
consonantal root. Within the data, medial glide root consonants sometimes surface as
a long vowels within either the lexeme’s singular or plural forms (such as the singular
sajjid becoming the plural saada ‘polite forms of address’ with the root consonants /sj-d/). Since zooraq’s weak medial root consonant, /w/, does not appear as a consonant
in the singular but rather as the long vowel /oo/, it uses the value of <stem sg 2> in the
qad_singular node to form a template. This template is structured by the coding in
(9) below.
(9) Subset of qad_singular node
QUAD_SINGULAR:
<stem sg 2> == ”<c 1>” ”<vowel sg>” ”<vowel sg>” ”<c 3>”
”<vowel sg 2>” ”<c 4>”.
As is necessary to output zooraq, the template is structured to place a long vowel after
the first root consonant and does not call for a <c 2>. However, in the plural template,
the glide value of <c 2> is required and therefore structured in qad_PL_inflc2’s <stem
pl 2> value creating z-a-w-a-a-r-i-q → zawaariq.
The structure of these lexical entries simulates my intended adherence to Kihm’s theoretical framework. Specifically, each stem forming node, such as inflc2, provides a
template for either a singular or plural form while the lexical entry defines the vowel
timbre and consonantal roots.7 The plural stem formation nodes are organized according
to the occurrence of the long vowel (or lack thereof as in dibb → dibab ‘bears’), with
further variation for stems within each node. An example of this organization can be
seen in (10) below.
7
The inheritance defined between a lexical entry and the V(owel) nodes should be viewed as simply a selection process from the set of vowels permitted for this particular language. The purpose of separating them
from the lexical entry itself was a foresight so the theory could be expanded further to encompass verbal
and other derivations.
229
Lindley Winchester
(10) inflc3 node coding
INFLC3:
<stem pl> == ”<c 1>” ”<vowel pl 2>” ”<c 2>” ”<c 3>” ”<vowel pl>”
”<vowel pl>” n
<stem pl 2> == ʔa ”<c 1>” ”<c 2>” ”<vowel pl 2>” ”<c 3>”
”<vowel pl>” ”<vowel pl>” ʔ.
In this example, we see the plural templates for those words with long vowel inflection
after the third root consonant. <stem pl> creates words such as ɤuraab ‘crow’ → ɤirbaan
‘crows’ while <stem pl 2> forms plurals such as sˤadiiq ‘friend’ → ʔasˤdiqaaʔ ‘friends’.
A simplistic hierarchical representation of bp formation as it is constructed in the
theory can be seen in Figure 4 below.
LexicalEntry
V
stemPL
NOINFL
INFLC1
INFLC2
INFLC3
Q-PL-NOINFL
Q-PL-INFLC2
dibab
saada
gibaal
ɤirbaan
sˤajadla
tazaalik
Figure 4: Inheritance network for plural formation.
Working from the bottom of the tree, examples of bps from Table 3 are located under
their corresponding stem formation nodes. From left to right we have a class for bps
with no apparent long vowel inflection (noinfl), with long vowel inflection following
the first consonantal root (inflc1), following the second consonantal root (inflc2), and
following the third consonantal root (inflc3). The two classes located at the far right of
the tree are designed for quadriconsonantal roots. These are further divided by whether
the quadriconsonantal bp shows long vowel inflection after the second root consonant
(q-pl-inflc2) or not at all (q-pl-noinfl). Altogether, these plural stem formation nodes
represent the fourteen distinct bp forms seen in Table 3 and exemplify the variation
found in bps across the language.
5 Conclusions
In an attempt to construct a wide array of complex broken plural forms in Egyptian
Arabic, the fundamentals of Kihm’s Root-and-Site Hypothesis can be integrated into
datr. Though encountering difficulties within the theoretical framework for portions of
230
14 Egyptian Arabic broken plurals in DATR
the data, the theory generates exemplary singular and plural forms for each of the designated inflection classes and subclasses into which the data has been organized. Therefore,
it covers the extent of complex variation found within the data set through an extension
of the theoretical framework. In this analysis, it has been shown that not only theoretical
but computational approaches can be utilized in the representation of complex morphological phenomena like the broken plural.
References
Badawi, El-Said & Martin Hinds. 1986. A dictionary of Egyptian Arabic. Beirut: Librairie
Du Liban.
Brown, Dunstan & Andrew Hippisley. 2012. Network morphology: A defaults-based theory
of word structure. Vol. 133. Cambridge: Cambridge University Press.
Evans, Roger & Gerald Gazdar. 1989. Inference in DATR. In Proceedings of the Fourth
Conference on European Chapter of the Association for Computational Linguistics, 66–
71. Association for Computational Linguistics.
Evans, Roger & Gerald Gazdar. 1996. DATR: A language for lexical knowledge representation. Computational Linguistics 22(2). 167–216.
Gadalla, Hassan AH. 2000. Comparative morphology of standard and egyptian arabic.
Vol. 5. Lincom Europa Munich.
Hammond, Michael. 1988. Templatic transfer in Arabic broken plurals. Natural Language
and Linguistic Theory 6(2). 247–270.
Kihm, Alain. 2006. Nonsegmental concatenation: A study of Classical Arabic broken
plurals and verbal nouns. Morphology 16. 69–105.
Lewis, M. Paul (ed.). 2009. Ethnologue: Languages of the World, Sixteenth edition. Dallas:
SIL International.
McCarthy, John J. & Alan S. Prince. 1990. Foot and word in prosodic morphology: The
Arabic broken plural. Natural Language and Linguistic Theory 8.2. 209–283.
Pollard, Carl & Ivan A. Sag. 1994. Head-driven phrase structure grammar. Chicago: University of Chicago Press & Stanford: CSLI Publications.
Ratcliffe, Robert R. 1998. The broken plural problem in Arabic and comparative semitic:
Allomorphy and analogy in non-concatenative morphology. Amsterdam/Philadelphia:
John Benjamins.
Ratcliffe, Robert R. 2013. Morphology. In Jonathan Owens (ed.), The Oxford handbook of
morphology, 71–91. Oxford: Oxford University Press.
231
Part IV
Syntax
Chapter 15
Factive relative clauses in Pulaar
Ibrahima Ba
The University of Kansas
Drawing from Kayne (1994a), this chapter shows that Headed Relative Clauses and Factive
Clauses in Pulaar are built from similar structures. Both display word order similarities, and
in each case the complementizer, which is homophonous with the determiner, agrees with
the (null or overt) head NP in Spec,CP. The verb form is also the same in Headed Relatives
and Factive Relatives, and it undergoes the same agreement pattern. Furthermore, Headed
Relatives and Factives in Pulaar both exhibit island constraints such that extraction out of
either construction is impossible; this indicates that they all involve movement of some sort.
The difference between these constructions is that the Headed Relative has an overt head
noun whereas Factives have null head nouns.
1 Introduction
This paper investigates factive relative clauses in Pulaar, a West Atlantic language spoken in Senegal and other West African countries. The Pulaar variety described here is
spoken in the southern part of Senegal. Specifically, the paper provides an analysis of
two factive constructions in Pulaar, namely the verbal factive and the ko-factive, as (1a)
and (1b) respectively: in (1a), the infinitive form of the verb is fronted and followed by the
complementizer; in (1b), the particle ko1 (glossed as a relative complementizer) always
appears at the leftmost edge of the clause.2
(1)
a. [def-go ngo ndef-mi ñebbe ngo] bettu Hawaa.
cook-inf c.rel cook-1sg beans cl.the surprise Hawaa
‘The fact that I cooked beans surprised Hawaa.’
‘The cooking that I cooked the beans surprised Hawaa.’
b. [ko ndef-mi ñebbe ko] bettu Hawaa.
c.rel cook-1sg beans cl.the surprise Hawaa
‘The fact that I cooked beans surprised Hawaa.’
Verbal Factive3
ko-Factive
1
Ko has a variety of meanings in Pulaar, most of which are not related semantically. I treat these various
instances of ko as homophones, which have meanings/functions such as focus/topic (see Cover 2006),
copula, noun class, complementizer, pronoun.
2 The two meanings of (1a) are discussed in §3.2.
Ibrahima Ba. Factive relative clauses in Pulaar. In Jason Kandybowicz, Travis Major, Harold
Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected papers from
the 45th Annual Conference on African Linguistics, 235–251. Berlin: Language Science Press.
DOI:10.5281/zenodo.1251736
Ibrahima Ba
The main claim in this paper is that the constructions in (1) are relative clause constructions with a derivation similar to headed relative clauses in Pulaar, as in (2):
(2)
Musa ñaam-ma [ñebbe ɗe ndef-mi ɗe].
Musa eat-perf beans c.rel cook-1sg cl.the
‘Musa ate the beans that I cooked.’
I argue that headed relatives as well as factive relatives can be derived from the same
underlying structure in (3) following Kayne (1994b). The structure in (3) is composed of
a D and a CP complement.
(3)
DP
D’
D
CP
C’
C
TP
This is explicitly shown in the structures in (4) where we can see the different movement operations that occur in the derivation of the different clauses. Specifically, the
entire CP moves to Spec, DP.
(4)
a. Verbal Factive
DP
D’
D
ngo
CP
defgo
C
ngo
C’
TP
ndef-mi ñebbe
236
15 Factive relative clauses in Pulaar
b. Ko-Factive
DP
c. Headed RC
DP
D’
D
ko
D’
CP
∅
C
ko
D
ɗe
C’
CP
ñebbe
TP
C
ɗe
ndef-mi ñebbe
C’
TP
ndef-mi t
The remainder of this paper is structured as follows: §2 provides a short background
on Pulaar which will include the basic word order, some properties of the noun and the
agreement morphology. The distribution of factive clauses is laid out in §3. §4 deals with
the structural similarities that exist between Headed Relatives and Factives in Pulaar.
§5 demonstrates that both headed relatives and factives are islands and §6 shows the
derivation of Headed Relatives and Factive clauses. §6 presents concluding remarks.
2 Background on Pulaar
Lewis (2009) states that Pulaar belongs to Atlantic branch of the Niger-Congo language
family. There is a large number of Pulaar dialects with varying levels of mutual intelligibility, spoken from Senegal to Cameroon and Sudan and all the countries in-between.
There are at least four dialects of Pulaar in Senegal: Futa Tooro region (north-east),
Fula(kunda) spoken in the Kolda region (south), Pular (spelled with one ‘a’ ) spoken
by people originally from Guinea Republic; and the dialect spoken in Kabaadaa (south
and east of Kolda), also known as Toore, which this paper is based on.
2.1 Word order
Pulaar is used here as a general term to refer to the language. It is a Subject-Verb-Object
(SVO), prepositional language, as shown in the sentence below.
(5) Taalibe mo jangu-m
deft-are nde les
lekki.
student cl.the read-perf.neut book-cl cl.the under tree
‘The student has read the book under a tree.’
237
Ibrahima Ba
Focus in Pulaar is generally encoded by the particle ko which precedes the focused
phrase, as shown in the example below:
(6)
a. (Ko) raandu ndu Musaa yii-noo.
foc dog.cl cl.the Musaa see-past
‘It’s the dog that Musaa saw.’
b. Musaa (ko) yii-no raandu ndu.
Musaa foc see-past dog.cl the.cl
‘Musaa saw the dog (not heard it bark).’
DP focus
Verb focus
The parentheses indicate that ko is optional. In the absence of ko, focus can still be
interpreted from the verb ending. Long vowels indicate DP focus whereas short vowel
indicate Verb focus, regardless of the presence or absence of of the focus particle ko. Ko
is also used in Wh-questions, as in the following example:
(7)
Ko Musaa yii-noo?
what Musaa see-past
‘What did Musaa see?’
Wh-question
2.2 Nouns in Pulaar
Pulaar is a noun class language. It has twenty-two noun classes and the noun class
marker follows the noun (Sylla 1982: 34).
(8)
a. raa-ndu ndu
dog-cl cl.the
‘the dog’
b. daa-ɗi ɗi
dog-cl cl.the
‘the dogs’
The noun in (8a) can be analyzed as the root noun raa “dog” and a suffix ndu. Thus,
the noun always occurs as a combination of the noun and the suffix, like raandu “a dog”.
The infinitive in Pulaar is composed of the verb root and the infinitive suffix go, as seen
in the examples in (9a-b). This infinitive form occurs in a variety of positions within a
sentence. The examples below show the different positions that the infinitive can occupy.
(9)
238
a. Mbiɗo yiɗi/foti
def-go maaro.
1sg
want/should cook-inf rice
‘I want to cook rice.’
b. O ñoot-ma tuuba am ba
ñoot-go wesoo.
3sg sew-perf pants my cl.the sew-inf beautiful
‘He has sewn my pants a beautiful sewing.’
Complement of V
As a noun + adjective
15 Factive relative clauses in Pulaar
(9b) shows that the infinitive in Pulaar can be modified by an adjective, which suggests
that it behaves as a noun belonging to the ngo class. Table 1 shows the noun classes in
Pulaar.
Table 1: Noun Classes in Pulaar.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Noun class
example
gloss
mo
nde
ndi
ndu
nge
ngo
ngu
nga
ba
ka
ki
ko
ɗum
ɗam
nge
ka
ngi
nga
ɓe
ɗe
ɗi
koñ
suko mo
hoore nde
ngaari ndi
raandu ndu
nagge nge
jungo ngo
pucuu ngu
damnga nga
mbabba ba
laanaa ka
leɓii ki
huuko ko
ɓaleejum ɗum
ndiyam ɗam
laacee nge
leyka ka
damngii ngi
neɗɗaa nga
yimɓe ɓe
gite ɗe
babaaji ɗi
laanoñ koñ
the child
the head
the ox
the dog
the cow
the hand
the horse
the door
the donkey
the plane, boat
the knife
the grass
the black thing
the water
the little tail
the small land
the huge door
the huge person
the people
the eyes
the donkeys
the small boats
Noun classes 1 to 18 are singular and noun classes 19 to 22 are plural. The noun class
1 is used for humans and borrowed words. It has two plural forms: 19 for humans and
21 for borrowed words. However, while 19 relates specifically to humans, 21 is not only
related to borrowed words; it is also the plural of other noun classes such as 3, 4, 5, 7
etc. The noun class 20 is also the plural of several noun classes such 8, 10, 2, etc. The
noun class 22 is the plural for diminutives 15 and 16. The augmentative classes 17 and 18,
however, have the regular plural class 20 even when the “augmented” noun denotes a
human referent.
For the remainder of this paper, I will be spelling nouns as one single unit, for instance
raandu instead of a split word raa-ndu.
239
Ibrahima Ba
Table 2: Singular/Plural Mapping of Noun Classes.
Singular
Plural
mo
ndi, ndu, nge, ngu, ba, ko, ɗum, ɗam
nde, ngo, ka, ki & the augmentatives nga, ngi
nge, ka (diminutives)
ɓe (humans), ɗi (loanwords)
ɗi
ɗe
koñ
2.3 Consonant mutation
Consonant mutation refers to the change of one consonant into another under certain
conditions. According to Sylla (1982) and McLaughlin (2005), Pulaar exhibits consonant
mutation, for instance the alternation between y, g and s, c below:
(10)
a. yitare ‘eye’
b. gite ‘eyes’
(11)
a. sengo ‘side’
b. cengle ‘sides’
Table 3 shows the alternation patterns that can be found in Pulaar.
Table 3: Mutating Initial Consonants.
Initial consonant of the verb
Simple
Mutated
Ø4 , g
f
h
b, w
s
j, y
d, r
ŋg
p
k
mb
c
ñj
nd
Alternations like these occur in a variety of contexts such as subject agreement on the
verb, singular/plural alternation on nouns, but also affixation. In what follows, I show an
example of each of these alternations. In matrix clauses for instance, subject agreement
is shown on the verb through the mutation of the initial consonant when the subject is
plural.
4 The symbol ‘Ø’ represents cases when the verb starts with a vowel. In such cases, [ŋg] becomes the mutated
sound in the right context.
240
15 Factive relative clauses in Pulaar
(12)
a. mi/a/o
sood-ma
oto.
I/you/he/she buy-perf.neut car
‘I/you have bought a car.’
b. En/on/ɓe
cood-ma
oto.
We/you/they buy-perf.neut car
‘We have bought a car.’
Singular
Plural
In (13a) the sentence has a singular subject and the verb ‘buy’ starts with [s]. In (13b),
however, where the subject is plural the verb ‘buy’ begins with <c> and is pronounced
[ʧ].
Consonant mutation may also occur in nominalization; that is when a verb is turned
into a noun, as shown in the following examples:
(13) Verb to Noun Alternations
a. surku-go → curki ‘smoke’
smoke-inf
‘to smoke’
b. yim-go → jimoo ‘a song’
sing-inf
‘to sing’
Note the alternations in examples (13a) and (13b) in which the initial consonant of the
verb changes in the corresponding noun.
3 Distribution and semantic interpretations of factives
3.1 Distribution of factives
Both factive clause types occur as subjects and complements to factive predicates, i.e.
predicates that presuppose the truth of their subjects or complements. For instance, the
sentence in (14), from Kiparsky & Kiparsky (1970), involves the non-factive verb ‘claim’.
In other words, a claim may be proven either right or wrong, as shown in (14b-c):
(14)
a. John claims that he offended Mary.
b. … and in fact, he did.
c. … but in fact, he did not.
Non-factive Predicate
The example in (15), however, involves a factive verb. That means it refers to an event
that has necessarily occurred, as shown in (15b-c):
(15)
a. John regrets that he offended Mary.
b. … and in fact, he did.
c. # … but in fact, he did not.
Factive Predicate
241
Ibrahima Ba
The examples in (16b) and (16c) respectively show verbal and ko factives as subjects:
(16)
a. ɓe nguju-m deftare.
(input to (16b-c)
3.pl steal-perf book
‘They stole a book.’
deftare ngo] bettu-mii-m.
Verbal-Factive
b. [wuju-go ngo ɓe nguj-i
steal-inf c.rel 3.pl steal-perf book cl.the surprise-1sg-perf
‘The fact that they stole the book surprised me.’
c. [ko ɓe nguj-i
deftare ko] bettu-mii-m.
ko-Factive
c.rel 3.pl steal-perf book cl.the surprise-1sg-perf
‘(The fact) that they stole the book surprised me.’
In Pulaar, factive clauses occur as arguments of factive verbs like bettugo ‘surprise’,
loɓgo ‘to be angry’, ricitaago ‘to regret’. Factive clauses can, thus, be complements to
factive verbs, as in the following examples where the verbal and the ko factive are objects
of the verb ricitaago ‘to regret’:
(17)
a. ɓe ndicit-iim [wuju-go ngo ɓe nguj-i
deftare ngo].
1pl regret-perf steal-inf c.rel 3.pl steal-perf book cl.the
‘They regret the fact that they stole the book.’
b. ɓe ndicit-iim [ko ɓe nguj-i
deftare ko].
1pl regret-perf c.rel 3.pl steal-perf book cl.the
‘They regret (the fact) that they stole the book.’
Verbal-Factive
ko-Factive
Also, factive clauses do not occur as arguments of non-factive verbs like siɓ-go ‘to
doubt’, as shown in the following examples:
(18)
a. * Verbal-Factive
deftare ngo].
mbiɗo siɓ-i
[wuju-go ngo ɓe nguj-i
1sg
doubt-perf steal-inf c.rel 3.pl steal-perf book cl.the
Intended: ‘I doubt the fact that they stole a book.’
b. * ko-Factive
mbiɗo siɓ-i
[ko ɓe nguj-i
deftare ko].
1sg
doubt-perf c.rel 3.pl steal-perf book cl.the
Intended: ‘I doubt that they stole a book.’
3.2 Semantic interpretations of Pulaar factive clauses
There are interpretive differences between the verbal factive and the ko-factive in Pulaar. In fact, whereas the verbal factive is ambiguous between an eventive and a manner
readings, the ko-factive can be interpreted as an event.
242
15 Factive relative clauses in Pulaar
(19)
a. [def-go ngo ndef-mi ñebbe ngo] bettu Hawaa.
cook-inf c.rel cook-1sg beans cl.the surprise Hawaa
‘The fact that I cooked beans surprised Hawaa.’
‘The cooking that I cooked the beans surprised Hawaa.’
b. [ko ndef-mi ñebbe ko] bettu Hawaa
c.rel cook-1sg beans cl.the surprise Hawaa
‘The fact that I cooked beans surprised Hawaa.’
Verbal-Factive
ko-Factive
The example in (19a) can mean that Hawaa did not expect the speaker to cook the
beans in the first place; maybe they agreed that the beans were for sale. In addition to
this eventive reading, the verbal factive has a manner reading under which (19a) would
mean that Hawaa expected the speaker to cook the beans but the cooking turned out to
be either so good or so bad that Hawaa is somehow surprised.
As for the ko-factive, it only has an eventive reading. In (19b) for instance, Hawaa is
surprised that the speaker cooked the beans. There may be a few reasons to this; Hawaa
may not have expected or wanted the beans to be cooked or she may not have expected
or wanted the speaker to cook the beans he/she does not like cooking or is a terrible
cook, etc.
4 Pulaar relative clauses
In this section I show the morphological similarities between factive clauses and headed
relative clauses. Specifically, I show that factive clauses are types of relative clauses. In
addition to being head initial, these three constructions have agreeing complementizer,
final determiner, similar placement for subject DP or pronoun. They also have the same
agreement properties.
4.1 Clause structure of headed relative clauses
Pulaar has head-initial relative clauses. The relativizer (or complementizer) agrees with
and follows the head noun. It is homophonous with the clausal determiner at the end of
the clause which encodes definiteness. When it is omitted, the head noun is indefinite.
The relative complementizer is obligatory.
(20)
mo
a. simis mo Hawaa loot-i
shirt c.rel Hawaa wash-perf cl.the
‘the shirt that Hawaa washed’
b. simis *(mo) Hawaa loot-i
shirt c.rel Hawaa wash-perf
‘(some) shirt that Hawaa washed’
Headed Relative Clause
243
Ibrahima Ba
(21)
ngo
a. faɗoo ngo Hawaa watt-ii
shoe c.rel Hawaa wear-perf cl.the
‘The shoe that Hawaa is wearing’
b. faɗoo *(ngo) Hawaa watt-ii
shoe c.rel Hawaa wash-perf
‘(some) shoe that Hawaa is wearing’
The examples in 20 have all the same material, the only difference is that (20a) ends
with a determiner which is missing in (20b). However, the complementizer in (20b) cannot be deleted. The same can be said (21) where the only difference is that (21b) is lacking
the final determiner; and again the complementizer is mandatory.
Subject agreement is shown on the verb through consonant mutation for plural subjects, as in matrix clauses. This is shown in the examples below:
(22)
ɗe
a. ñebbe ɗe Hawaa def-i
beans c.rel Hawaa cook-perf cl.the
‘the beans that Hawaa cooked’
b. ñebbe ɗe ndef-mi ɗe
beans c.rel cook-1sg cl.the
‘the beans that I cooked’
ndef-i
ɗe
c. ñebbe ɗe rewɓe ɓe
beans c.rel women cl.the cook-perf cl.the
‘the beans that the women cooked’
3sg subject
1sg subject
3pl subject
The initial consonant of the verb changes from [d] in (22a) to [nd] in (22b,c). DP subjects in relative clauses always precede the verb.
Table 4: Pulaar subject pronouns.
Singular
Plural
mi
a
o
min, en
on
ɓe
The word order of the headed object relative clauses in Pulaar is as follows:
(23)
244
NP
c.rel
S
V
Otrace
det.cl
15 Factive relative clauses in Pulaar
4.2 Clause Structure of factive clauses
Verbal factives are so called because a form of the verb (the infinitive or gerundive) is
treated as a noun heading the factive clause. In this clause, the nominalized form of the
verb is followed by an agreeing relativizer which is homophonous with the determiner
at the end of the clause. This can be seen in the examples below:
(24)
loot-go ngo Hawaa loot-i
wutte ngo
wash-inf c.rel Hawaa wash-perf shirt cl.the
‘the fact that Hawaa washed a shirt’
(25)
wutte ko
ko Hawaa loot-i
c.rel Hawaa wash-perf shirt cl.the
‘(the fact) that Hawaa washed a shirt’
Verbal Factive
When the determiner is omitted, the verbal noun is indefinite5 . The relative complementizer is obligatory. This is shown in the following examples:
(26)
wutte
Loot-go *(ngo) Hawaa loot-i
wash-inf c.rel Hawaa wash-perf shirt
‘A/some washing that Hawaa washed a shirt’
(27)
wutte6
*ko Hawaa loot-i
c.rel Hawaa wash-perf shirt
‘the fact that Hawaa washed a shirt’
In verbal factive constructions, the verb appears to show some form of agreement.
Subject agreement is shown on verb through consonant mutation for plural subjects, as
in matrix clauses. However, singular subjects also trigger consonant mutation when they
follow the verb. This is shown in the examples below:
(28)
ñebbe ngo
a. def-go ngo Hawaa def-i
cook-inf c.rel Hawaa cook-perf beans cl.the
‘the fact that Hawaa cooked beans’
b. def-go ngo ndef-mi ñebbe ngo
cook-inf c.rel cook-1sg beans cl.the
‘the fact that I cooked beans’
ndef-i
ñebbe ngo
c. def-go ngo ɓe
cook-inf c.rel subj.pro cook-perf beans cl.the
‘the fact that they cooked beans’
3sg subject
1sg subject
3pl subject
5 This is still interpreted as a factive. Structures like (26) and (27) can be answers to a question like: ‘What is so
and so mad about’ where the person answering the question is not making it sound like their interlocutor
knew about that specific event.
6
This is just interpreted as a subject focus construction and means something along the lines: ‘It’s Hawaa
who cooked/washed…’.
245
Ibrahima Ba
The initial consonant of the main clause verb changes from [d] in (28a) to [nd] in
(28b,c). DP subjects in relative clauses always precede the verb, as in (28a). However, all
subject pronouns, except 3sg/pl, have to follow the verb. In this case, the initial consonant of the verb mutates even when the subject pronoun is singular, as in (28b).
The word order in a verbal factive appears to be the following:
(29)
VNom
c.rel
S
V
O
det.cl
I assume that the infinitive form of the relative verb (VNOM ) is moved to Spec,CP to
fill in for a null noun ‘fact’ (which does not exist in Pulaar) along the lines of Collins
(1994) and Tamba & Torrence (2013). Assuming that only the verb root has been moved,
the presence of the infinitive suffix can be justified by the need for agreement; VNOM ,
the complementizer and the determiner must all agree.
4.3 Clause structure of the ko-factive
With ko as a relativizer, the ko-factive is headless, or it is rather headed by a null noun.
This is due to the fact that Pulaar does not have the word ‘fact’. But one piece of evidence is also that this null noun is associated with an existing noun class ko. When
the determiner is omitted, the structure cannot be interpreted as a factive. The relative
complementizer is obligatory. This is shown in the following examples:
(30) *ko Hawaa loot-i wutte7
c.rel Hawaa wash-perf shirt
‘the fact that Hawaa washed a shirt’
(31)
*ko Hawaa def-i
ñebbe
c.rel Hawaa cook-perf beans
‘the fact that Jeyla cooked beans’
Similar to verbal factive and headed relative constructions, the verb show of agreement morphology in ko-factives. Subject agreement is shown on verb through consonant
mutation for plural subjects, as in matrix clauses. This is shown in the examples below:
(32)
7
a. ko Hawaa def-i
ñebbe ko
c.rel Hawaa cook-perf beans cl.the
‘the fact that Hawaa cooked beans’
b. ko ndef-mi ñebbe ko
c.rel cook-sg beans cl.the
‘the fact that I cooked beans’
3sg subject
1sg subject
This is just interpreted as a subject focus construction and means something along the lines: ‘It’s Hawaa
who cooked/washed…’.
246
15 Factive relative clauses in Pulaar
3pl subject
ndef-i
ñebbe ko
c. ko ɓe
c.rel 3rd .pl cook-perf beans cl.the
‘the fact that they cooked beans’
The initial consonant of the verb changes from [d] in (32a) to [nd] in (32b,c). DP subjects always precede the verb. However, all subject pronouns, except 3sg/pl, have to
follow the verb. In this case, the initial consonant of the verb mutates even when the
subject pronoun is singular, as seen (32b).
The word order in a ko-factive appears to be the following:
(33)
ØNP
c.rel
S
V
O
det.cl
Based on the data presented here, the headed relative clause and factive relative clauses
share a similar structural pattern, as shown below:
(34)
a.
b.
c.
NP
VNP
ØNP
C.REL
C.REL
C.REL
S
S
S
V
V
V
Otrace
O
O
DET.CL
DET.CL
DET.CL
Headed relative
Verbal factive
ko-factive
Factive clauses involve a null noun for the ko-factive and a verb with nominal features
for the verbal factive and both of these nominals agree with a specific complementizer
and the corresponding homophonous determiner or noun class. I assume the presence
of a null noun in the ko-factive due to the fact that it agrees with a noun class, but also
there is no noun ‘fact’ in Pulaar.
The clear parallel that exist between the headed relative clause and factive relative
clauses suggest that these constructions look like NP [CP] Det. I will follow Kayne
(1994b) and analyze relative clauses as involving a D + CP, as in the structure in (35):
(35)
DP
D’
D
CP
DP/NPi
C-rel
C’
TP
247
Ibrahima Ba
However, whether these constructions are all derivable from the same structure is
dependent upon whether or not they all involve some type of movement.
The data below suggest that relativization and factivization involve movement. In fact,
relativization or ‘factivization’ out of a relative clause is impossible in headed relatives
as well as the verbal and ko-factive clauses. The examples below illustrate this fact:
(36)
a.
ɗa yiɗ-i
[suko mo Isa tott-i
ñebbe mo.]
2sg like-perf child clrel Isa give-perf beans cl.the
‘I like the boy that Isa gave beans’
mo].
RC
b. * ɗa yiɗ-i
[ñebbe ɗe [suko mo Isa tott-i
2sg like-perf [beans clrel [child clrel Isa give-perf] cl.the
‘You like the beans that boy that Isa gave’
ñebbe mo ngo.
[tottu-go ngo [suko mo Isa tott-i]
c. * ɗa yiɗ-I
2sg like-perf [give-inf clrel [child clrel Isa give-perf] beans cl.the cl.the
VF
ñebbe ɗe.
mo ko
[ko [suko mo Isa tott-i]
d. * ɗa yiɗ-i
2sg like-perf [clrel [child clrel Isa give-perf] cl.the cl.the beans cl.the
Ko-F
The examples in (36b-d) show that it is impossible to relativize (or ‘factivize’) out of a
relative clause. The examples (36b), (36c) and (36d) show, respectively, a relative clause,
a verbal factive and a ko-factive. The impossibility of extracting out of a relative clause
or relativizing out of a relative clause indicates that these constructions involve some
type of movement and are islands.
5 Derivation of relative and factive clauses
In this section, I provide a unified analysis of RCs and factive clauses. Following Tamba
& Torrence (2013), Torrence (2005) and Kayne (1994b), I assume that in Pulaar, headed
relatives and factives can be derived from the same underlying structure which consists
of a D and a CP complement. I argue that in this structure CP raises to Spec,DP.
I first analyze relative clauses like (35):
(37)
ti mo
wuttei mo Hawaa loot-i
shirt cl.rel Hawaa wash-perf cl.the
‘The shirt that Hawaa washed’
Headed Relative Clause
In this construction, the head (object) NP moves to Spec,CP as shown in (38):
248
15 Factive relative clauses in Pulaar
(38)
DP
D’
D
mo
cl.the
CP
DP/NPi
wutte
shirt
C’
C
mo
C-rel
TP
Hawaa loot-i t
Hawaa washed
In the second step of the derivation, CP moves to Spec,DP to yield the surface structure,
as it appears in (39).
Turning to verbal factives, along the lines of Tamba & Torrence (2013) and following
Collins (1994) and Aboh (2005), I argue that in the Verbal Factive in (39a), a copy of
the verb, which is relativized and carries the infinitival –go, is moved to Spec,CP. The
complementizer agrees in noun class with the infinitival verb in Spec,CP. As have I have
pointed out, the infinitive form the Pulaar verb exhibits nominal properties8 .
(39)
wutte ngo
a. loot-go ngo Hawaa loot-i
Wash-inf c.rel Hawaa wash-perf shirt cl.the
‘the fact that Hawaa washed a shirt’
b.
DP
D’
D
ngo
cl.the
8
CP
DP/NPi
loot-go9
wash-inf
C’
C
ngo
C-rel
TP
Hawaa loot-i wutte
Hawaa washed shirt
See example (9b).
249
Ibrahima Ba
Once the infinitival verb has moved to Spec,CP, the whole CP node is then moved to
Spec,DP generating the expected surface structure.
This analysis correctly derives the word order of the Verbal Factive construction in
(39a) in a way similar to the derivation of the headed relative.
I now move to the ko-factive structure. The ko-Factive Relative is slightly different
from the other relative types because it involves a null NP meaning ‘fact’. But the presence of this null NP is signaled by its agreement with some noun class, in this case ko.
In order to derive a ko-Factive like the one in (40a), we can posit the movement of the
null NP from inside the TP to Spec,CP. As a second step, the movement of CP to Spec,DP
yields the surface word order along the lines of Headed Relatives and Verbal Factives, as
we can see in (40b):
(40)
wutte ko
a. ko Jeyla loot-i
c.rel Jeyla wash-perf shirt cl.the
‘(the fact) that Jeyla washed a shirt’
b.
DP
D’
D
ko
cl.the
CP
C’
DP/NPi
∅np
C
ko
C-rel
TP
Hawaa loot-i wutte
Hawaa washed shirt
As the analysis has shown, Headed Relatives and Factive Relatives in Pulaar can all
be derived from the same hierarchical structure in a relatively similar manner.
6 Concluding remarks
In this paper, I have argued that Headed Relatives and Factive Relatives have similar
structure in sense that they have a similar word order and in all of them the complementizer agrees with the (null or overt) head NP in Spec,CP and is homophonous with the
determiner.
9A
reviewer notes that the fact the verb copy is infinitival indicates that there is more structure involved. I
leave for future research the precise nature of the nominal constituent in Spec,CP and how a verb becomes
nominalized.
250
15 Factive relative clauses in Pulaar
In my analysis, the differences between the three types has to do with the material
in Spec,CP. In headed RCs, it is a lexical noun. In the verbal factives, it is a nominalized
copy of the verb, while in the ko-factives it is a null noun of the ko class.
References
Aboh, Enoch O. 2005. Deriving relative and factive contructions in Kwa. In Laura Brugè,
Giuliana Giusti, Nicola Munaro, Walter Schweikert & Giuseppina Turano (eds.), Contributions to the Thirtieth Incontro di Grammatica Generativa. Cafoscarina, 265–285.
Venice: Libreria Editrice Cafoscarina.
Collins, Chris. 1994. The factive construction in Kwa. New York: Cornell University. (Unpublished manuscript).
Cover, Rebecca. 2006. Focus on ko: The syntax, semantics, and pragmatics of identificational focus in Pulaar. Berkeley: University of California Berkeley dissertation.
Kayne, Richard S. 1994a. The antisymmetry of syntax. Cambridge, Massachusetts: MIT
Press.
Kayne, Richard S. 1994b. The antisymmetry of syntax. Cambridge, MA: MIT Press.
Kiparsky, Paul & Carol Kiparsky. 1970. Fact. In Manfred Bierwisch & Karl Erich Heidolph
(eds.), Progress in linguistics, 143–173. The Hague: Mouton.
Lewis, M. Paul (ed.). 2009. Ethnologue: Languages of the World, Sixteenth edition. Dallas:
SIL International.
McLaughlin, Fiona. 2005. Reduplication and consonant mutation in the Northern Atlantic languages. In Bernhard Hurch (ed.), Studies on reduplication, 111–133. Berlin:
Mouton de Gruyter.
Sylla, Yèro. 1982. Grammaire moderne du Pulaar. Dakar, Sénégal: Nouvelles Éditions
Africaines.
Tamba, Khady & Harold Torrence. 2013. Factive relative clauses in Wolof. Annual Meeting
of the Linguistic Society of America handout.
Torrence, Harold. 2005. A promotion analysis of relative clauses in Wolof. Annual Meeting of the Berkeley Linguistics Society 31(2). 107–118.
251
Chapter 16
Object suffixes as incorporated
pronouns in Seereer
Nico Baier
In Seereer (Atlantic, Senegal), singular pronominal objects are obligatorily marked by an object suffix on the verb. This paper provides the first comprehensive description of this object
suffixation pattern, a topic that has been only cursorily described in the extant literature on
Seereer (cf. Renaudier 2012). In addition, I provide a preliminary theoretical account of the
Seereer object suffix system. I argue that Seereer object suffixes are best analyzed as incorporated pronouns. Evidence for such an analysis comes from the following: (i) an object suffix
may never occur with an in situ object DP; (ii) an object suffix may not double an extracted
object in relative clauses, wh-questions, or focus constructions; (iii) there is only one object
suffix allowed per clause; and (iv) an object suffix may reference either object in a double
object construction. I argue that object suffixes raise to Spec-vP and are subsequently incorporated in the verb via m-merger (Matushansky 2006, Kramer 2014, Harizanov 2014). This
analysis elegantly derives the behaviors listed above. Such an approach also allows us to
integrate the Seereer object suffixation data into the broader understanding of cliticization
patterns crosslinguistically, thereby enriching our understanding of object marking systems
in verbs.
1 Introduction
In Seereer (Atlantic; Senegal), singular object pronouns are marked by a suffix on the
verb, as shown in (1). Plural object pronouns are realized as a full pronominal DP (2).1
(1)
Singular Object Suffixes
✄
a. Jegaan a naf-a-✂xam ✁.
Jegaan 3 hit-dv-1sg.obj
‘Jegaan hit me.’
✄
b. Jegaan a naf-a- ang .
✂
✁
Jegaan 3 hit-dv-2sg.obj
‘Jegaan hit you.’
✄
c. Jegaan a naf-a-✂an ✁.
Jegaan 3 hit-dv-3sg.obj
‘Jegaan hit him/her/it.’
1 Plural object pronouns are preceded by the differential object marker a. This marker is required with objects
that are pronouns or proper names. I will not discuss the differential object marker here.
Nico Baier. Object suffixes as incorporated pronouns in Seereer. In Jason Kandybowicz, Travis
Major, Harold Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected
papers from the 45th Annual Conference on African Linguistics, 253–268. Berlin: Language Science
Press. DOI:10.5281/zenodo.1251738
Nico Baier
(2)
Plural Object Pronouns
a. Jegaan a naf-a a in.
Jegaan 3 hit-dv obj 1pl
‘Jegaan hit us.’
b. Jegaan a naf-a a nuun.
Jegaan 3 hit-dv obj 2pl
‘Jegaan hit you guys.’
c. Jegaan a naf-a a den.
Jegaan 3 hit-dv obj 3pl
‘Jegaan hit them.’
There are only singular object suffixes; no equivalent plural object suffixes exist in the
language. Alongside the suffixes, Seereer has a full set of free pronouns for all person/
number combinations. The object suffixes and the free pronouns are shown below in
Table 1:
Table 1: Object Suffixes vs. Free Pronouns
Object Suffix
Free Pronoun
1sg
2sg
3sg
1pl
2pl
3pl
-aam
mi
-ong
wo’
-in
ten(o)
in
nun
den(o)
In Table 1, the object suffixes are given in their underlying forms. In most cases, these
underlying forms are obscured by morphonological processes. For reasons of space I will
not discuss these processes here.2
Although there is a small amount of published work on Seereer (Faye 1982; McLaughlin 1994; 2000; Renaudier 2012), there is no comprehensive description of the object suffix
system. This paper aims to fill this gap. I show that object suffixes are best analyzed as
pronouns that are morphologically incorporated into the verb, rather than object agreement. I also sketch a preliminary analysis of the pronoun incorporation process. Building
on analyses of pronominal clitics by Harizanov (2014) and Kramer (2014), I propose that
object suffixes originate in an argument position as pronouns and undergo head movement to v 0 .
The structure of this paper is as follows. In §2, I show that the distribution of object suffixes is identical to the distributio of free object pronouns, and argue that this
shows object suffixes to be incorporated pronouns. I then discuss constraints on object
suffixation in §3. Based on these facts, I present my analysis in §4. Section §5 provides
conclusions.
2 Though
see Renaudier (2012) for discussion of the morphonology of object suffixes in a different Seereer
dialect, Seereer-Marlodj.
254
16 Object suffixes as incorporated pronouns in Seereer
2 Object suffixes are pronouns
In this section, I show that object suffixes have the same distribution as other object
pronouns and are therefore best analyzed as incorporated pronouns. Evidence for this
comes from the fact that object suffixes cannot double an in situ object DP; that they
cannot co-occur with an Ā-moved object; and that they must resume a left-dislocated
object.
2.1 Doubling of full NPs
In situ full DP objects can never co-occur with a coreferential object suffix on the verb,
as shown by the pair of examples in (3a-b):
(3)
a. Mataar a jaw-a [DP maalo fe]
Mataar 3 cook-dv
rice det
‘Mataar cooked the rice.’
✄
[DP maalo fe]i
b. * Mataar a jaw-a- ani
✂
✁
rice det
Mataar 3 cook-dv-3sg.obj
Intended: ‘Mataar cooked the rice.’
In (3a), there is a single full, post-verbal full DP object, maalo fe ‘the rice’. When a object
suffix coreferential with maalo fe is added to the verb in (3b), the sentence becomes
ungrammatical. Seereer is completely invariant with respect to this constraint. As shown
in (4), an object suffix can never double any kind of full DP object:
(4)
[DP ya’
um oxe]i .
a. * Jegaan a bug-a-ani
mother 3poss det
Jegaan 3 love-dv-3sg.obj
Intended: ‘Mataar loves his mother.’
[DP a mi]i .
b. * Jegaan a ga’-a-xami
obj 1sg
Jegaan 3 see-dv-1sg.obj
Intended: ‘Jegaan saw me.’
[DP okoor oxe]i .
c. * Jegaan a ga’-a-ani
man det
Jegaan 3 see-dv-3sg.obj
Intended: ‘Jegaan saw the man.’
[DP muus ne]i .
d. * Jegaan a ga’-a-ani
cat det
Jegaan 3 see-dv-3sg.obj
Intended: ‘Jegaan saw the cat.’
[DP mbin ne]i
e. * Jegaan a jik-a-ani
house det
Jegaan 3 buy-dv-3sg.obj
Intended: ‘Jegaan bought the house.’
Kinship term
Free pronoun
Human animate
Non-human animate
Inanimate
255
Nico Baier
So, the basic observation is that full, post-verbal DP objects are in complementary distribution with object suffixes. This observation is immediately explained if we assume
that object suffixes and full DP objects occupy the same structural position at some
point in the derivation. Thus, object suffixes and full DP objects compete for an argument position, as there can only be one argument per structural position. This, in turn,
straightforwardly follows if we assume that object suffixes are pronouns that have been
incorporated morphologically into the verb.
2.2 Object extraction contexts
Object suffixes are also in complementary distribution with an Ā-extracted object. This
is true for all constructions that involve Ā-extraction in Seereer: wh-questions, focus
clauses, and relative clauses.3 First, an object suffix cannot co-occur with an extracted
object wh-phrase, as shown in (5a-b):
✄
(5) a. * xari Ami a jik-u- ni ?
✂ ✁
what Ami 3 buy-ext-3sg.obj
Intended: ‘What did Ami buy?’
Inanimate wh-word
✄
b. * ani Ami a bug-u- ni ?
✂ ✁
who Ami 3 love-ext-3sg.obj
Intended: ‘Who does Ami love?’
Animate wh-word
This constraint is also active in object focus clauses, as shown in (6a-b):
✄
(6) a. * Jegaani foc Ami a bug-u- ni .
✂ ✁
Jegaan
Ami 3 love-ext-3sg.obj
Intended: ‘It’s Jegaan that Ami loves.’
DP focus
✄
b. * (a) wo’i foc Ami a bug- ongi .
✂
✁
Ami 3 love-2sg.obj.ext
obj 2sg
Intended: ‘It’s you that Ami loves.’
Pronoun focus
Finally, in object relative clauses, an object suffix may not double the extracted DP, as
seen in (7):
✄
Ami a ñam-uu- ni -a]
(7) * [N maalo]i [CP ne
✂ ✁
rice
rel.det Ami 3 eat-ext-3sg.obj-rel
Intended: ‘the rice that Ami ate’
The data in (5)-(7) also follow from the idea that object suffixes are underlyingly pronouns
that saturate argument positions. An Ā-extracted argument must be generated in an
3 Evidence
that these clauses involve Ā-extraction of the object comes from the fact that the verb takes the
final suffix -u, which only occurs when Ā-movement has occurred in a clause within which the verb is
contained. See Baier (2014) for extensive discussion.
256
16 Object suffixes as incorporated pronouns in Seereer
argument position before it undergoes Ā-movement, and this blocks an object suffix
from being generated in the same argument position. Note that plural object pronouns,
which do not have a suffixal form, are also blocked from co-occuring with an extracted
plural DP object:
(8)
a. * aniini Ami a bug-u a deni ?
who.pl Ami 3 love-ext obj 3pl
Intended: ‘Who all does Ami love?’
b. * (a) nuuni foc Ami a ga’-u a nuuni .
obj 2pl
Ami 3 see-ext obj 2pl
Intended: ‘It’s you all that Ami saw.’
Plural wh-word
Plural DP focus
So object suffixes have the exact same distribution as free, plural object pronouns in
cases of object Ā-extraction. This is further evidence that object suffixes are pronouns
that are incorporated into the verb.
2.3 Left dislocation contexts
Object suffixes must double a left-dislocated full DP object. As shown in (9), when the
dislocated DP is singular, an object suffix is required on the verb:
✄
(9) a. maalo fe, Mataar a jaw-a-✂an ✁.
rice det Mataar 3 cook-dv-3sg.obj
‘The rice, Mataar cooked it.’✄
Suffix
b. * maalo fe, Mataar a jaw-a-✂Ø ✁.
rice det Mataar 3 cook-dv
Intended: ‘The rice, Mataar cooked it.’
No Suffix
Free singular object pronouns may also be dislocated. Resumption by an object suffix is
also required in this case:
(10)
a. (a) mi, Mataar a bug-a-*(xam).
obj 1sg Mataar 3 cook-dv-1sg.obj
Intended: ‘Me, Mataar likes.’
b. (a) wo’, Mataar a bug-a-*(ang).
obj 2sg Mataar 3 cook-dv-1sg.obj
Intended: ‘You, Mataar likes.’
Again, the behavior of object suffixes is the same as that of free plural object pronouns.
When a plural object DP is left dislocated, a plural pronoun is required as a resumptive,
(11a); lack of one results in ungrammaticality (11b):
✄
(11) a. goor we, Mataar a ga’-a a ✂den ✁.
men det Mataar 3 see-dv obj 3pl
‘The men, Mataar saw them.’
Pronoun
257
Nico Baier
✄
b. * goor we, Mataar a ga’-a ✂Ø ✁.
men det Mataar 3 see-dv
Intended: ‘The men, Mataar saw them.’
No Pronoun
Left dislocation in Seereer does not involve Ā-movement. Evidence for this comes from
the fact that left dislocation does not trigger the presence of the Ā-sensitive final suffix
-u.4 Instead, left dislocation involves base generation of a DP in the left periphery and
resumption in an argument position in the main part of the clause. Since resumptive
elements are usually pronouns (McCloskey 2006), this supports the idea that object suffixes are themselves pronouns. Again, this idea is reinforced by the fact that they pattern
identically to free plural pronouns in this construction.
3 Syntactic constraints on object suffixation
In the previous section, I presented distributional evidence that object suffixes are in fact
pronouns that end up as a morphological subunit of the verb word. Following this line
of thought, I assume that, as pronouns, object suffixes are generated as D heads in object
position as the complement to V. This is shown in (12), where ‘OS’ stands for object suffix:
(12)
[VP V [D OS ]]
Thus, object suffixes are simply generated in argument position like any other object
and later become associated morphologically with the verb. But why do object suffixes
incorporate into the verb? In this section, I present evidence that object suffixation is
constrained by the syntactic structure of the clause and therefore object suffixation is a
fundamentally syntactic process. The specific data are derived from the following contexts:
(13)
a. The obligatoriness of object suffixes
b. Multiple object constructions: Ditransitives, applicatives, causatives
c. Object suffixes in passive clauses
3.1 Obligatoriness
If there is only one singular object pronoun, it must always surface as a suffix, never as
a free pronoun, as shown by (14).
✄
(14) a. Jegaan a fal-a- ang .
✂
✁
Jegaan 3 kick-dv-2sg.obj
‘Jegaan kicked you.’
Object suffix
4
For further discussion, see Baier (2014).
258
16 Object suffixes as incorporated pronouns in Seereer
✄
b. * Jegaan a fal-a
(a) ✂wo’ ✁.
Jegaan 3 kick-dv obj 2sg
Intended: ‘Jegaan kicked you.’
Free pronoun
Regardless of the presence of other post-verbal constituents, a singular object pronoun
must be realized as a suffix. Consider (15), which shows that a free singular object pronoun is impossible in such contexts:
✄
(15) a. Jegaan a fal-a- ang
faak.
✂
✁
Jegaan 3 kick-dv-2sg.obj yesterday
‘Jegaan kicked you yesterday.’
b. * Jegaan a fal-a
faak
(a) wo’.
Jegaan 3 kick-dv yesterday obj 2sg
Intended: ‘Jegaan kicked you yesterday.’
c. * Jegaan a fal-a
(a) wo’ faak.
Jegaan 3 kick-dv obj 2sg yesterday
Intended: ‘Jegaan kicked you yesterday.’
Note that, otherwise, objects are generally freely ordered with regards to other postverbal constituents. As shown in (16), plural object pronouns and full DP objects may
precede or follow an adverb such as faak ‘yesterday’:
(16)
a. Jegaan a ga’-a (a nuun) faak
(a nuun).
Jegaan 3 kick-dv obj 2pl yesterday obj 2pl
‘Jegaan saw you guys yesterday.’
(otew oxe).
b. Jegaan a ga’-a (otew oxe) faak
Jegaan 3 kick-dv woman det yesterday woman det
‘Jegaan saw the woman yesterday.’
Plural pronoun
Full DP
These data are important in that they show that object suffixation is insensitive to linear
order. If object were sensitive to linear order, we would expect a clause like (14b), in
which an adverbial intervenes between a singular object pronoun and the verb, to be
grammatical (as the plural counterpart in (16a) is). However, this order is not possible.
Since syntactic operations not sensitive to linear order, this points to a syntactic account
of object suffixation.
3.2 Multiple object constructions
Seereer has several types of double object constructions (DOC). Such constructions occur
with lexical ditransitive verbs, such as ci’ ‘give’; verbs bearing one of the applicative
suffixes -an ‘benefactive’ and -(i)t ‘instrumental/locative’; and causative verbs derived
with the causative suffix -noor. Lexical ditransitive verbs and applicative verbs pattern
259
Nico Baier
together with regard to word order and object suffixation, while causative verbs pattern
differently than the first two classes with regard to these diagnostics.
Ditransitive verbs and applicative verbs in Seereer are symmetrical double object constructions (following the terminology of Bresnan & Moshi 1990). When ditransitive and
applicative verbs have two full DP arguments and both are post-verbal, these arguments
are freely ordered. This is shown for ditransitives in (17) and for the benefacative applicative -an in (18). In the following examples, ‘↔’ indicates that the bracketed constituents
can be reversed in order:
(17)
a. Jegaan a ci’-a
[DP okoor oxe]goal ↔ [DP atere le]theme .
book det
Jegaan 3 give-dv
man det
‘Jegaan gave the man the book.’
✓ goal < theme / ✓ theme < goal
(18)
a. a jaw-an-a
[DP okoor oxe]ben ↔ [DP maalo fe]theme .
rice det
3 cook-ben-dv
man det
‘He cooked the rice for the man.’
✓ ben < theme / ✓ theme < ben
When one of the objects of a ditransitive or applicative verb is a singular pronoun, it
must be realized as a suffix, as shown for a ditransitive verb in (19).5 This constraint
holds regardless of order, as shown by (19b-c):
✄
(19) a. Jegaan a ci’-a- anggoal [DP atere le]theme .
✂
✁
book det
Jegaan 3 give-dv-2sg.obj
‘Jegaan gave you the book.’
Object suffix
[DP a wo’]goal [DP atere le]theme .
b. * Jegaan a ci’-a
book det
obj 2sg
Jegaan 3 give-dv
Intended: ‘Jegaan gave the book to you.’
Free pronoun
c. * Jegaan a ci’-a
[DP atere le]theme [DP a wo’]goal .
obj 2sg
Jegaan 3 give-dv
book det
Intended: ‘Jegaan gave the book to you.’
Free pronoun
When a ditransitive or applicative verb takes two singular object pronouns, either argument may surface as a suffix, as shown in (20a-b).6 However, there is a maximum of one
object suffix per verb form; the verb cannot take multiple object suffixes, as shown by
(20c):
✄
(20) a. Jegaan a ci’-a- anggoal [DP a ten]theme .
✂
✁
Jegaan 3 give-dv-2sg.obj
obj 3sg
‘Jegaan gave you it.’
Goal suffix
5 For
reasons of space, I will use data only from lexical ditransitives for the remainder of this section. The
judgements also apply to all applicatives.
6 In cases where one object is a speech act participant and the other is not, my consultant showed a preference
for suffixation of the SAP object. However, this is not a hard and fast constraint. Examples like (20a) are
perfectly grammatical.
260
16 Object suffixes as incorporated pronouns in Seereer
✄
b. Jegaan a ci’-a- antheme [DP a wo’]goal .
✂
✁
Jegaan 3 give-dv-3sg.obj
obj 2sg
‘Jegaan gave it to you.’
☎
✞
c. * Jegaan a ci’-a- anggoal -intheme .
✝
✆
Jegaan 3 give-dv-2sg.obj-3sg.obj
Intended: ‘Jegaan gave you it.’
Theme suffix
Two suffixes
So these particular multiple object constructions are symmetrical with regard to object
suffixation, in that either object may be realized as an object suffix when they are both
singular pronouns.
On the other hand, causatives of transitive verbs derived with the suffix -noor are
asymmetrical double object constructions. Such verbs take two objects: the subject of
the caused event (the causee) and the underlying object of the caused event. With regard
to word order, a full DP causee must always precede a full DP object, as shown in (21):
(21)
a. Jegaan a fal-noor-a
[DP okoor oxe]causee [DP naak le]object .
cow det
Jegaan 3 kick-caus-dv
man det
‘Jegaan made the man kick the cow.’
✓causee < object
b. * Jegaan a fal-noor-a
[DP naak le]object [DP okoor oxe]causee .
man det
Jegaan 3 kick-caus-dv
cow det
Intended: ‘Jegaan made the man kick the cow.’
*object < causee
This is the opposite of what we saw for ditransitive and applicative verbs, where either
ordering was licit. Also unlike ditransitive and applicative verbs, there is an asymmetry
for causative verbs with regards to which argument is able to appear as an object suffix.
The causee must be an object suffix if it is a singular pronoun, as shown by (22):
✄
(22) a. Jegaan a fal-noor-a- angcausee [DP naak le]object .
✂
✁
cow det
Jegaan 3 kick-caus-dv-2sg.obj
‘Jegaan made you kick the cow.’
Object suffix
b. * Jegaan a fal-noor-a
[DP a wo’]causee [DP naak le]object .
cow det
Jegaan 3 kick-caus-dv
obj 2sg
‘Jegaan made you kick the the cow.’
Free pronoun
However, the object of the causative verb cannot be realized as an object suffix, even if
it is the only singular object pronoun in the clause, as shown by (23a):
(23)
a. Jegaan a fal-noor-a
[DP okoor oxe]causee [DP a wo’ ]object .
obj 2sg
Jegaan 3 kick-caus-dv
man det
‘Jegaan made the man kick the cow.’
Free pronoun
✞
☎
b. * Jegaan a fal-noor-a- angobject [DP okoor oxe]causee .
✝
✆
Jegaan 3 kick-caus-dv-2sg.obj
man det
‘Jegaan made the man kick you.’
Object suffix
261
Nico Baier
Again, this is exactly the opposite of what we saw with ditransitives and applicatives.
Like those verbs, however, it is also impossible for a causative verb to take two object
suffixes, as shown by (24):
☎
✞
(24) * Jegaan a fal-noor-a- angcausee -inobject .
✝
✆
Jegaan 3 kick-caus-dv-2sg.obj-3sg.obj
Intended: ‘Jegaan made you kick it.’
Two suffixes
All of the facts just discussed are summarized in Table 2:
Table 2: Sereer double object constructions
Type
Word Order
Object Suffix
Multiple Suffixes
Ditransitive
Applicative
Causative
sym
sym
asym
sym
sym
asym
✗
✗
✗
The differences between symmetrical (ditransitive/applicative) and asymmetrical (causative)
double object constructions are a convincing argument in favor of a syntactic account of
object suffixation. As we will see below, these differences can be relativized to independent principles of locality in which causatives include a barrier to object suffixation of
the internal argument of the causativized predicate, whereas ditransitives and applicatives do not.7 A non-syntactic account would have to stipulate these differences.
In addition, the general ban on multiple suffixes is an argument against approaches
to object suffixation that do not take place in the syntax, as such accounts would have
to posit a different set of weak pronouns that occur as suffixes, and a stipulation would
be required to block these suffixes from co-occuring. A syntactic approach, on the other
hand, can take advantage of the idea that the operation triggering incorporation of a
pronoun into the verb only applies once per structure.
3.3 Passives
The final constraint on object suffixation concerns passives. When a ditransitive verb is
passivized, one of the underlying objects is promoted to subject, while the other object
is left behind in the post-verbal position and treated as an object. Either object may be
promoted to subject, as shown in (25):
(25)
7
[DP atere le]theme
a. [DP okoor oxe]goal a ci’-e
3 give-pass
book det
man det
‘The man was given the book.’
See Baker et al. (2012) for such an approach to similar data in Lubukusu.
262
Goal subject
16 Object suffixes as incorporated pronouns in Seereer
[DP okoor oxe]goal
b. [DP atere le]theme a ci’-e
3 give-pass
man det
book det
‘The book was given to the man.’
Theme subject
In (25a), the goal argument is promoted to subject and the theme remains post-verbal
as an object. In (25b), the theme is promoted to subject and the goal argument remains
behind. When the object that remains post-verbal is a singular pronoun, it cannot be
realized as a suffix. This is true regardless of which argument it refers to, as shown by
(26):
✄
(26) a. * [DP okoor oxe]goal a ci’-e-✂n ✁
3 give-pass-3sg.obj
man det
Intended: ’The man was given it.’
Goal suffix
✄
b. * [DP atere le]theme a ci’-e-✂n ✁
3 give-pass-3sg.obj
book det
Intended: ‘The book was given to him/her.’
Theme suffix
In (26a), the object suffix on the verb corresponds to the theme argument. In (26b), the
object suffix on the verb refers to the goal argument. Both examples are ungrammatical. This ungrammaticality is avoided by realizing the pronominal object as a full, free
pronoun.
(27)
[DP a ten]theme
a. [DP okoor oxe]goal a ci’-e
3 give-pass
obj 3sg
man det
‘The man was given it.’
b. [DP atere le]theme a ci’-e
[DP a ten]goal
book det
3 give-pass
obj 3sg
‘The book was given to him/her.’
As seen in (27), a post-verbal object in a ditransitive passive is grammatical, while a object suffix is not. This observation is another argument for a syntactic approach to object
suffixation, as we expect different voice types to enforce different syntactic constraints.
An account that locates object suffixation in a post-syntactic module of the grammar
would have to appeal to a stipulation by stating that singular pronouns cannot be realized as suffixes in a structure with a passive. Alternatively, one could say that there is a
templatic restriction banning incorporation into a passive verb. A syntactic analysis, on
the other hand, can appeal to differences in the structure of active and passive sentences
to account for the availability of object suffixation. For instance, perhaps object suffixation is triggered by a head present in the active that is not present in the passive. I now
move on to sketching such an approach in section 4.
263
Nico Baier
4 Towards an analysis
Before moving on to my analysis, I present a summary of the generalizations made above
concerning object suffixation in (28):
(28)
Characteristics of Object Suffix
a. There are only singular object suffixes
b. An object suffix may not co-occur with an in situ DP.
c. An object suffix may not co-occur with an Ā-extracted object (focus/wh-relative)
d. An object must co-occur with a topicalized object.
e. There is a limit of one object suffix per verb.
f. An object suffix is obligatory where possible.
g. An object suffix may refer to either argument in a symmetrical DOC.
h. An object suffix cannot refer to the theme of a causativized transitive verb.
i. An object suffix cannot occur in a passive verb.
In this section, I sketch an analysis that aims to capture the generalizations given above.
The core idea of my analysis is that object suffixation involves head movement of a
pronoun (D0 ) to the head v 0 , which causes it to be morphologically incorporated into
the verb. This idea is schematized in (29):
(29)
[vP V+v+OS [VP V [D OS ]]]
There are two questions that must be answered with regards to the structure in (29). First,
what triggers movement of a pronominal D0 to v 0 and why does it only target singular
pronouns? Second, why is the head movement impossible in some circumstances, such
as when there are multiple objects or when the verb is passive?
Building on analyses of Bulgarian pronominal clitics by Harizanov (2014) and Amharic
object suffixes by Kramer (2014), I suggest that incorporation of a pronoun into v 0 is
motivated by the operation Agree which is triggered by a probe on v 0 . Both Harizanov
and Kramer and adopt the conception of head movement developed by Matushansky
(2006) in which head movement is taken to be regular phrasal movement to a specifier
followed by a special operation m-merger which fuses a specifier with its head. They
argue that clitic doubling in Amharic and Bulgarian derives from movement of a DP to
specifier of v, after which the DP m-merges with v 0 . For Harizanov, m-merger of a XP
reduces that projection to its label, yielding a complex head. This is shown in (30):
(30)
a. [vP DP [ v [VP V DP ]]]
move
b. [vP D+v [VP V DP P ]]]
DP moves to Spec-vP
M-Merger of DP
Under this analysis, object suffixation in Seereer occurs because v 0 is equipped with a
probe that causes a pronoun to move to its specifier. Later, that pronoun undergoes mmerger with v 0 , resulting in morphological incorporation of the pronoun into the verb.
264
16 Object suffixes as incorporated pronouns in Seereer
I propose that active v 0 in Seereer is equipped with a number probe ([u#]) that triggers
movement of an argument in VP to Spec-vP. I follow much work on the operation Agree
in assuming that probes can be relativized to search for specific values of a feature (Béjar
2008; Béjar & Rezac 2009; Preminger 2011). In this case, I assume that the number probe
on v 0 is relativized to search for singular features. I represent this as [u#sg ].
Assuming that the #-probe on v 0 is relativized to search only for singular features
immediately derives the fact that only singular pronouns will incorporate into the verb
in Seereer, yielding only singular object suffixes. But how do we derive the fact that no
doubling of an in situ DP object is possible in Seereer? Recall that the head movement
approach I am employing assumes that DPs can undergo m-merger to form a complex
head with v 0 . Thus, clitic doubling should, in principle, be possible.
I propose that the ability for XPs to undergo m-merger is subject to parametric variation. In languages like Bulgarian it is possible, and therefore clitic doubling occurs. In
languages like Seereer, however, it is not possible, and therefore DPs can never be doubled by object suffixes, as these suffixes are impossible to generate. Thus, we have two
situations in Seereer, given in (31) and (32).
(31) Singular Pronoun = m-merger
vP
VP
v
D
v+[u#SG ]
V
DSG
(32) Singular DP = no m-merger
vP
DP
v+[u#SG ]
VP
V
DSG
In (31), the complement of V is a singular pronoun, a minimal D0 , and therefore, object
suffixation occurs. In (32), on the other hand, the complement of V is a singular DP.
Therefore, m-merger of DP is not possible after it moves to Spec-vP and no object suffix
surfaces. This derives the fact that there is no doubling of full DPs by object suffixes in
Seereer.
A key characteristic of object suffixation in Seereer is that it is obligatory when it is
possible, but when it is impossible, no ungrammaticality results. This is problematic for
265
Nico Baier
the idea that suffixation is triggered by Agree, as we would expect sentences without
singular objects and active v to be ungrammatical. To alleviate this problem, I follow
Preminger (2011) in assuming that the failure of a probe to find matching features does
not result in crash. Therefore, a #-probe can be present on every active v, but derivations
without a singular DP object will not crash. This derives the generalization that object
suffixes are obligatory when there is a singular object pronoun, but when there is not
one, the sentence is fine.8
Furthermore, because there is only one #-probe on v 0 , only one object suffix is possible
on any given verb. Thus, I assume that once the #-probe on v 0 has found a matching
singular DP, it does not have to probe further, and is satisfied. Thus, when there are two
singular object pronouns in the structure, as in a DOC, the higher object pronoun in the
structure is found by the #-probe on v 0 , and that pronoun incorporates. The other is left
free:
(33)
[vP [ v [VP Di [ V Dk ]]]]
DP moves to Spec-vP
In (33), the #-probe on v 0 finds the higher of two object pronouns, and thus that one is
the only one that is incorporated.
Finally, this analysis is able to derive two further constraints on object suffixation.
First, because v is responsible for encoding the voice of the clause, it is reasonable to
assume that the #-probe is limited to certain v heads. Namely, passive v lacks the #probe, and therefore, no object suffix is possible in passive structures. Second, the differences between symmetrical DOCs (ditransitives/applicatives) and asymmetrical DOCs
(causatives) can be derived by appealing to Phase-based locality (Chomsky 2001; 2008).
In causative DOCs, there is a phase boundary between the causee object and the theme
object which blocks Agree with the theme. In symmetrical DOCs, on the other hand,
there is no such boundary, and therefore both objects can occur as suffixes.
5 Conclusion
In this paper, I have presented a description of Seereer object suffixes, focusing on their
distribution and syntactic behavior. On the basis of their distributional characteristics,
I have argued that they are best analyzed as pronouns that are morphologically incorporated into the verb. I have further argued that this process of incorporation occurs in
the syntax, in that it is constrained by syntactic structure. These constraints include the
fact that object suffixation is obligatory; that it cannot occur more than once per verb;
that it is sensitive to the voice of the clause; and that it is sensitive to the structure of
double object constructions. I have also also sketched an implementation of the syntactic
approach based on the idea that active v 0 in Seereer bears a #-probe relativized to search
for singular DPs, and that this probe triggers head movement of pronouns to adjoin to v 0 .
8
An alternative would be to posit that the probe is only sometimes present on v 0 . However, pursuing this
line of thinking would require one to devise a way to enfore the probe’s presence when there is at least
one singular object pronoun in the structure. I will avoid this discussion here, and leave the comparison of
the two analyses to future work.
266
16 Object suffixes as incorporated pronouns in Seereer
Acknowledgments
I thank Peter Jenks and Line Mikkelsen for insightful comments, guidance, and discussion on this project, as well as the audience at ACAL 45 at the University of Kansas. I am
also indebted to my consultant Malick Loum for taking the time to share his knowledge
of Seereer with me. All data in this paper were gathered during the 2012–2013 UC Berkeley Field Methods class and subsequent follow-up work with John Merrill in 2013–2014.
Abbreviations
det
dv
ext
inf
obj
pl
determiner
default vowel
extraction suffix
infinitive
object
plural
rel
sg
1
2
3
relative
singular
first person
second person
third person
References
Baier, Nico. 2014. Spell-out, Chains, and Long Distance Wh-movement in Sereer. Ms., UC
Berkeley.
Baker, Mark C., Ken Safir & Justine Sikuku. 2012. Sources of (a)symmetry in Bantu double
object constructions. MS: Rutgers.
Béjar, Susana. 2008. Conditions on Phi-Agree. In Daniel Harbour, David Adger & Susana
Béjar (eds.), Phi theory, 130–154. Oxford: Oxford University Press.
Béjar, Susana & Milan Rezac. 2009. Cyclic agree. Linguistic Inquiry 40(1). 35–73.
Bresnan, Joan & Lioba Moshi. 1990. Object asymmetries in comparative Bantu syntax.
Linguistic Inquiry 22(2). 147–186.
Chomsky, Noam. 2001. Derivation by phase. In Michael Kenstowicz (ed.), Ken Hale: A
life in linguistics, 1–52. Cambridge, Massachusetts: MIT Press.
Chomsky, Noam. 2008. On phases. In Robert Freidin, Carlos Peregrın
́ Otero & Maria
Luisa Zubizarreta (eds.), Foundational issues in linguistic theory: Essays in honor of jeanroger vergnaud (Current Studies in Linguistics), 133–166. Cambridge, MA: MIT Press.
Faye, Souleymane. 1982. Morphologie du verbe sérère. Dakar: Centre de Linguistique Appliquée de Dakar.
Harizanov, Boris. 2014. Clitic doubling at the syntax-morphophonology interface: Amovement and morphological merger in Bulgarian. Natural Language and Linguistic
Theory 32. 1033–1088.
Kramer, Ruth. 2014. Clitic doubling or object agreement: The view from Amharic. Natural
Language and Linguistic Theory 32. 593–634.
Matushansky, Ora. 2006. Head movement in linguistic theory. Linguistic Inquiry 37(1).
69–109.
267
Nico Baier
McCloskey, James. 2006. Resumption. In Martin Everaert & Henk van Riemsdijk (eds.),
Resumption, vol. IV (The Blackwell Companion to Syntax), 94–117. Blackwell.
McLaughlin, Fiona. 1994. Noun classification in Seereer-siin. UT Austin dissertation.
McLaughlin, Fiona. 2000. Consonant mutation and reduplication in Seereer-siin. Phonology 17. 333–363.
Preminger, Omer. 2011. Agreement as a fallible operation. MIT dissertation.
Renaudier, Marie. 2012. Dérivation et valence en Seereer. Université Lumière Lyon 2 dissertation.
268
Chapter 17
Searching high and low for focus in
Ibibio
Philip T. Duncan
University of Kansas
Travis Major
University of California, Los Angeles
Mfon Udoinyang
University of Kansas
This paper discusses two strategies in Ibibio for focusing verbs: contrastive verb focus and
exhaustive verb focus. We demonstrate how these constructions differ crucially in the syntactic configurations and derivations that underlie each. Exhaustive verb focus is marked
by the presence of the focus operator kpɔ́t ’only’, which is base-generated high in the left
periphery and triggers phrasal movement of the TP containing the focused verb via piedpiping. Contrastive verb focus is marked by verb doubling produced by head movement, and
it invokes a low focus phrase situated in the middle field, somewhere at the boundary of the
inflectional and verbal domains. Both types of verb focus in Ibibio are thus syntacticallydriven, but the locus of each is split across the clausal spine, and each Foc head can probe
independent of the other. Ibibio thus furnishes further evidence that multiple foci can occur
in a single clause, and it also provides independent support for the existence of a low focus
phrase.
1 Introduction
In this paper, we discuss morphosyntactic properties of two types of focus involving
verbs in Ibibio: contrastive verb focus (1b) and exhaustive verb focus (1c).
(1)
a. ékpê á-mà á-kót
ŋ̀-wèt
ekpe 3sg-pst 3sg-read nmlz-write
‘Ekpe read a book.’
(input to 1b,c)
Philip T. Duncan, Travis Major & Mfon Udoinyang. Searching high and low for focus in Ibibio. In
Jason Kandybowicz, Travis Major, Harold Torrence & Philip T. Duncan (eds.), African linguistics
on the prairie: Selected papers from the 45th Annual Conference on African Linguistics, 269–288.
Berlin: Language Science Press. DOI:10.5281/zenodo.1251740
Philip T. Duncan, Travis Major & Mfon Udoinyang
b. ékpê á-ké
á-kòó-kót
ŋ̀-wèt
ekpe 3sg-pst.foc 3sg-con.foc-read nmlz-write
‘Ekpe READ the book (not, say, take it away).’
c. ékpê á-ké
á-kót
ŋ̀-wèt
kpɔ́t
ekpe 3sg-pst.foc 3sg-read nmlz-write only
‘Ekpe only read the book.’
Contrastive Verb Focus
Exhaustive Verb Focus
We motivate and explore two distinct focus positions corresponding to each construction, a high focus phrase (HFocP) in the C domain, and a low focus phrase (LFocP) in the
inflectional domain. We also show that Ibibio has both syntactically and semantically
distinct loci of verb focus. Exhaustive verb focus recruits structure high in the left periphery, and is derived by phrasal movement where the TP is pied-piped. On the other
hand, contrastively focused verbs are situated much closer to VP and are generated by
head movement, where V0 is attracted to the lower focus head. Moreover, we discuss
how these distinct structural configurations allow for double verb focus constructions.
The structural superiority of the phrasal projection that houses the exhaustively focused
verb triggers scope effects such that exhaustive focus takes wide scope over contrastive
focus obligatorily.
Ibibio thus provides independent evidence for multiple foci occurring in a single clause
(Krifka 1992; Rizzi 1997; Kiss 1998) and further support for the existence of a low focus
position (Belletti 2004). Our proposed analysis is given in (2), which shows the derivation
for both exhaustive verb focus and contrastive verb focus:
270
17 Searching high and low for focus in Ibibio
(2)
HFocP
HFoc’
HFoc
kpɔ́ t
CP
C
TP
DPSUBJ
probegoalEXFOC
T’
T
LFocP
DPSUBJ
LFoc’
LFoc
probegoalCONFOC
VP
DPSUBJ
V’
V
DPOBJ
This derivation illustrates our analyses for contrastive and exhaustive verb focus, and it
also demonstrates how both of these structurally distinct foci can be activated to generate double focus. In contrastive verb focus, LFoc0 probes V0 and triggers head raising. In
exhaustive focus, HFoc0 forms a probe-goal relation with V0 ; instead of generating head
movement, though, the TP is pied-piped to Spec,HFocP. When both foci are activated,
ordering is critical: contrastive verb focus must be embedded under exhaustive focus for
the derivation to be sustained.
This paper is organized as follows. In §2, we provide a basic background of Ibibio, focusing on word order and agreement, and motivating the existence of verb raising in
the language. Following this, in §3 we turn to argument focus and wh-questions to provide a backdrop for understanding verb focus constructions. §4–6 provide our analyses
of contrastive verb focus, exhaustive verb focus, and double verb focus, respectively. §7
concludes.
271
Philip T. Duncan, Travis Major & Mfon Udoinyang
2 Background
2.1 Word order & agreement
Ibibio is a Lower Cross Niger-Congo language spoken in Akwa Ibom state in southeastern Nigeria. It is an SVO language with both subject agreement and object agreement
(Essien 1990a; Baker & Willie 2010):
(3)
́
a. èkà
á-mà á-fát
áyɨn
mother 3sg-pst 3sg-hug child
‘The mother hugged the child.’
b. èkà
á-mà á-ɱ́-fát
(míèn)
mother 3sg-pst 3sg-1sg-hug 1sg
‘The mother hugged me.’
As seen in (3), subject agreement surfaces on both T0 and V0 , leading to multiple subject
agreement. Object agreement occurs on V0 only, and is not always visible in the surface
form.1
Ibibio matrix clauses project not only TP, but also a series of functional layers such as
AspP and MoodP. Subject agreement has “no firm upper limit” and is present on “every
verbal functional head” (Baker & Willie 2010: 110):
(4)
ú-nám
ú-màná-ké
ú-sé
ú-kpá
ú-ké
2sgS-cond 2sgS-perf 2sgS-impf 2sgS-do.again-neg 2sgS-do
‘You should not have been doing it again.’ (Baker & Willie 2010: 118)
These facts yield the following word order in a standard declarative clause:
(5)
DPsubj Agrsubj -T […] Agrsubj -/Agrobj -V DPobj
2.2 Verb raising
Verb movement occurs in several contexts in Ibibio. One of these is negation, illustrated
below:
(6)
1
a. òkón á-mà á-tóŋŋó
okon 3sg-pst 3sg-start
‘Okon had started.’
í-tóŋŋó-ké
b. òkón í-ké
okon 3sg-pst.foc 3sg-start-neg
‘Okon had not started.’
Affirmative
Negative
Oftentimes because of phonological reasons (e.g. vowel hiatus resolution) object agreement is difficult to
discern. All person markers in Ibibio are vowels except 1sg, which is a nasal that assimilates to the onset of
the verb root. Thus, object marking always survives in cases involving 1sg objects because the nasal does
not delete.
272
17 Searching high and low for focus in Ibibio
c. ímá á-mà á-dép
ima 3sg-pst 3sg-buy
‘Ima bought it.’
d. ímá í-ké
í-dép-pé
ima 3sg-pst.foc 3sg-buy-neg
‘Ima didn’t buy it.’
Affirmative
Negative
Note in these examples that negation surfaces as a CV suffix, which in these cases is
either -ké (6b) or an assimilated suffix (6d).2 These forms provide evidence for the order
of Tense and Negation, as well as morphosyntactic consequences of V raising (Baker &
Willie 2010).3,4 The abbreviated tree in (7) shows the formation of the complex head in
(6d):
(7)
T’
T0
NegP
í-ké
Neg0
í-dépj
-pé
VP
V0
tj
As in (7), NegP dominates VP, and V-to-Neg raising results in negation surfacing
postverbally.5
Verb raising also occurs in reciprocal constructions, which are bipartite in Ibibio, producing a suffix that resembles negation. Reciprocal morphology is circumfixal, as in (8c),
consisting of a du- prefix and a CV suffix6 :
2
There is also a third allomorph, -ɣV, which surfaces on monosyllabic verb roots. See Akinlabi & Urua
(2002), who also treat the various allomorphs of the negative suffix to be “underlyingly /ké/” (Akinlabi &
Urua 2002: 127).
3
In agreement with Baker & Willie (2010), we believe that V raising is supported by the fact that negation
surfaces preverbally as a separate word, ké, in small clause constructions (e.g. causatives) and subjunctives,
which may lack the TP layer. We remain agnostic at present with respect to the possibility of V raising
through Neg to T, though we feel this is a viable option (see Baker 2008).
4
We discuss below our account for the change in tense markers.
5
Alternatively, one reviewer points out, the negative suffix could result from V raising around Neg followed
by Neg encliticizing onto V (see Pollock 1989, for French; Holmberg & Platzack 1995, for Scandinavian). For
our purposes, though, either analysis predicts the same output, as we merely wish here to motivate the
existence of verb raising in Ibibio independent of contrastive verb focus.
6
As with the negative suffix, the reciprocal suffix form is assimilative and varies according to the syllable
structure and phonetic form of the verb root.
273
Philip T. Duncan, Travis Major & Mfon Udoinyang
a. é-mà é-kɨt́
3pl-pst 3pl-see
‘They saw.’
b. í-ké
í-kɨt́ -té
3pl-pst.foc 3sg-see-neg
‘They didn’t see.’
c. é-mà é-dû-kɨ̀t-tè
3pl-pst 3pl-rec-see-rec
‘They saw each other.’
(8)
Affirmative
Negation
Bipartite reciprocal
Negated reciprocals have stacked suffixes, as seen in (9), and negation appears farther
away from the verb than the reciprocal suffix:7
í-ké
í-dú-kɨ̀t-tè-kè
3pl-pst.foc 3pl-rec-see-rec-neg
‘They didn’t see each other’
(9)
Reciprocal + Negation
Ibibio verbs thus raise for structurally superior heads to surface postverbally (à la
Baker’s 1985 Mirror Principle). In (9), the bipartite reciprocal is formed prior to negation, and the ordering of the stacked suffixes gives insight into syntactic structure. The
schematic in (10) shows the derivation based on the hierarchy we posit to derive the
aforementioned properties of negatives and reciprocals:
7 The
semantics of negated reciprocals support our ordering where Neg >> Rec. Negation always takes
wide scope over the reciprocal, which suggests that the reciprocal verb constitutes the input to negation.
Additionally, negation can appear before a reciprocal verb in the effect clause of a causative:
(i)
eno á-mà á-nám
ɔ́mmɔ̂ ké í-dú-kɨ̀t-tè
eno 3sg-pst 3sg-make they neg 3pl-rec-see-rec
‘Eno made them not see each other.’
Following Baker & Willie (2010) we take it that the preverbal negative particle in (i) is the morphological
exponent of Neg0 when the verb does not raise. Note, though, that the verb bears reciprocal morphology,
which again suggests that the reciprocal suffix attaches to the verb before the negative suffix.
274
17 Searching high and low for focus in Ibibio
(10)
TP
T’
T0
NegP
Neg0
RecP
Rec0
VP
V0
Thus, if RecP intervenes between NegP and VP, V raising ensures that the reciprocal
suffix surfaces closest to the verb, since the verb head raises first to Rec0 . This forms a
complex head that provides the input to Neg0 . As we argue below, verb raising and the
architecture in (10) are significant for understanding contrastive verb focus, which also
involves head movement.
3 Argument focus & wh-questions
When arguments are focused, the past tense marker -mà is replaced with -ké:8
(11)
a. ànìyè (ówó) ké èkà
á-ké/*mà
á-fát
who person comp mother 3sg-pst.foc/*pst 3sg-hug
‘Who did the mother hug?’
́ ké èkà
b. (á-dò) áyɨn
á-ké/*mà
á-fát
3sg-be child comp mother 3sg-pst.foc/*pst 3sg-hug
‘It was the child that the mother hugged.’
Object wh-question
Object focus
Thus, in past tense, -mà is incompatible with argument focus (Essien 1990a,b; Willie &
Udoinyang 2012: 244). “Focus” -ké surfaces obligatorily in argument focus contexts (for
past tense), as well as wh-questions (Note that the 1st ké in (11a–b) is the complementizer;
the inflected á-ké—relevant for our discussion—is obligatory). Following Rizzi (1997), we
take it that the landing site of focused constituents and wh-expressions is a focus phrase
located in the C domain. In this paper, we call this projection HFocP to distinguish it
from a second focus phrase that we argue projects rather low in the clausal spine. HFoc0
bears a focus feature that draws a phrasal element to its specifier, presumably because
such movement is induced by the need to satisfy a focus-criterion (Rizzi 1997).
8 We
here only present data in the past tense, though present and future tenses pattern similarly in this
regard.
275
Philip T. Duncan, Travis Major & Mfon Udoinyang
In contrast to object wh-questions and object focus, an overt C is illicit in subject whquestions and subject focus. Moreover, past tense -mà cannot occur in these constructions, and the fact that -mà and “focus” -ké are in complementary distribution suggests
that the -ké in (12a, c) is “focus” -ké, not the complementizer.
(12)
ànìyé í-ké/*mà
í-fát
áyɨń
Subject wh-question
who 3sg-pst.foc/*pst 3sg-hug child
‘Who hugged the child?’
́
b. * ànìyé ké í-ké
í-fát
áyɨn
who comp 3sg-pst.foc 3sg-hug child
(Intended: ‘Who hugged the child?’ or ‘Who is it that hugged the child?’)
á-ké/*mà
á-fát
áyɨń
Subject focus
c. (á-dò) èkà
3sg-be mother 3sg-pst.foc/*pst 3sg-hug child
‘It was the mother that hugged the child (not the father).’
d. * èkà
ké á-ké
á-fát
áyɨń
mother comp 3sg-pst.foc 3sg-hug child
(Intended: ‘It was the mother that hugged the child [not the father].’)
a.
This subject-object asymmetry in argument focus suggests a “that-trace effect” (Perlmutter 1971; Chomsky & Lasnik 1977) disallowing subject extraction over overt complementizers.
In summary, argument focus in Ibibio requires a special tense marker (“focus” -ké in
past tense), and the neutral tense marker is illicit in such constructions. Focused arguments and wh-items undergo movement to HFocP in the complementizer domain, and
land higher than the C head. As we discuss below, these properties of focus constructions
are significant for differentiating between the two types of verb focus under consideration here: exhaustively focused verbs pattern much like argument focus constructions
and involve phrasal movement to the left periphery, whereas contrastively focused verbs
do not activate structure in the C system, and instead are derived in the inflectional domain via head movement.
4 Contrastive verb focus
4.1 Morphophonological properties
When verbs are contrastively focused, verb morphology expresses focus (Essien 1990a:
103–106; Akinlabi & Urua 2000; 2002; see Cook 2002 for verb focus in the closely related
Efik).
(13)
276
a. ákùn á-mà á-dép ŋ̀-wèt
akun 3sg-pst 3sg-buy nmlz-write
‘Akun bought the book.’
(input to 13b)
17 Searching high and low for focus in Ibibio
b. ákùn á-ké
á-dèé-dép
ŋ̀-wèt
í-ké
akun 3sg-pst.foc 3sg-con.foc-buy nmlz-write 3sg-pst.foc
í-yɨ̀p-pé-yɨ̀p
3sg.con.foc-neg-steal
‘Akun BOUGHT the book, she didn’t STEAL it.’
Forms of focused verbs demonstrate interactions between phonology, morphology,
and syntax. In affirmative forms, the focus component “takes the shape of a heavy (bimoraic) syllable” (Akinlabi & Urua 2002: 156), which appears on the surface to be some
type of prefixal “reduplicant.” Vowel lengthening occurs on the “reduplicant,” and the
initial CV sequence of the verb root becomes a “reduplicative prefix” of the form CVV-p.
This prefix bears a tone pattern (LH or HH) that is sensitive to the tone melody on the
root. The -ATR vowels /ɨ, ʉ, ʌ/ cannot be lengthened in Ibibio, and these change to [e,
u, ɔ] in order to be lengthened. Finally, verb roots with underlyingly low tones become
HL falling tones in contrastive reduplication. These properties can be seen in the examples of affirmative contrastively focused verbs in Table 1, which are given for each of the
vowels and simple tones in Ibibio.
Table 1: Contrastive verb focus forms
Vowel
(w/ tone)
Permissible
syllable type
Verb
English
gloss
[í]
[ì]
[ɨ]́
[ɨ̀]
[é]
[è]
[ú]
[ù]
[ʉ́]
[ʉ̀]
[ó]
[ò]
[ɔ́]
[ɔ̀]
[ʌ́]
[ʌ̀]
[á]
[à]
CV(C)
CV(C)
CVC
CVC
CV(C)
CV(C)
CV(C)
CV(C)
CVC
CVC
CV(C)
CV(C)
CVC
CVC
CVC
CVC
CV(C)
CV(C)
dí
kpì
́
tɨm
nɨ̀m
sé
wèt
túúk
fù
bʉn
bʉm
bót
bòn
tɔ́k
tɔ̀k
fʌk
tʌk
má
mà
‘come’
‘cut’
‘pound’
‘keep’
‘look’
‘write’
‘touch’
‘be lazy’
‘keep many things’
‘break’
‘mold’
‘begat’
‘urinate’
‘verbally abuse’
‘cover’
‘grate’
‘love’
‘complete’
Focused stem
(affirmative)
dìídí
kpìíkpî
́
tèétɨm
nèénɨ̂m
sèésé
wèéwêt
tùútúúk
fùúfû
bùúbʉ́n
bùúbʉ̂m
bòóbót
bòóbôn
tɔ̀ɔ́tɔ́k
tɔ̀ɔ́tɔ̂k
fɔ̀ɔ́fʌ́k
tɔ̀ɔ́tʌ̂k
màámá
màámâ
277
Philip T. Duncan, Travis Major & Mfon Udoinyang
4.2 Morphosyntactic structure
Unlike argument focus, which recruits structure in the C domain, we claim that the
derivation for verb focus is more local, that is, TP-internal:
(14)
ákùn [TP á-ké
á-yèé-yɨ̂p
ŋ̀-wèt
] í-ké
akun
3sg-pst.foc 3sg-con.foc-steal nmlz-write 3sg-pst.foc
í-dép-pé-dép
3sg-buy-neg-buy
‘Akun STOLE the book, she didn’t BUY it.’
Evidence for our claim comes from the position of contrastively focused verbs with
respect to T0 . We take it that the presence or absence of “focus” -ké is a diagnostic of
activation (or not) of the left periphery. Unlike argument focus, where “focus” -ké tense
marker appears obligatorily, contrastively focused verbs can occur with the standard
past tense -mà and without “focus” -ké:
(15)
́ mé
ímà á-mà á-ɲèé-ɲɨm
ima 3sg-pst 3sg-con.foc-agree
‘Ima AGREED (she didn’t disagree).’
Thus, contrastive verb focus does not activate the left edge. Instead, the focused verb
́ mé ‘AGREED’ in (15) surfaces below the T0 -mà.
ɲééɲɨm
To account for this, we posit a low focus projection that dominates VP, and propose
that verbs undergo movement to LFoc0 in contrastive verb focus. This is shown in the
abbreviated tree in (16), which shows the derivation of (15):
(16)
TP
T’
ímà
Ima
á-mà
3sg-pst
…
LFocP
LFoc0
́ mé i
á-ɲèé-ɲɨm
3sg-con.foc-agree
278
VP
ti
17 Searching high and low for focus in Ibibio
We argue that the contrastive verb focus “morpheme” is the product of the verb headmoving to LFoc0 (see (2) above), and that the syntax provides input to phonology, which
results in this special verb morphology. In the derivation in (16), LFoc0 probes for V0
(Chomsky 2000; 2001) and attracts it to itself. We take it that this probing and attraction
is driven by an interpretable focus feature on LFoc0 , which V0 values following head
adjunction. Focus “reduplication” is a post-syntactic consequence that results from head
raising. Interestingly, in Ibibio this low focus position is uniquely associated with contrastive semantics for verbs, which is sort of an unexpected restriction.9 We stipulate—
but leave for future investigation—that Ibibio LFoc0 has a property such that it probes
for features exclusive to verbs, and this disallows attracting phrasal units.
Negated verbs may offer insight into the syntactic structure of verb focus. As noted
above, V raising produces a CV negative suffix, as seen in (17b).
(17)
a. à-mà á-fɔ́p
2sg-pst 3sg-burn
‘You burned it.’
b. ú-ké
ú-fɔ́p-pɔ́
2sg-pst.foc 2sg-burn-neg
‘You didn’t burn it.’
Affirmative
Negative
Instead of a phonologically reduced copy of the verb that appears in affirmative contrastive verb focus forms, negative focused verbs exhibit two full copies of the verb (irrespective of syllable type) with Neg intervening:
(18)
a. ú-ké
ú-fɔ́p-pɔ́-fɔ́p
Neg + Contrastive Verb Focus
2sg-pst.foc 2sg-burn-neg-burn
‘You didn’t BURN it.’
b. í-ké
í-dép-pé-dép
ŋ̀-wèt
á-ké
á-yèé-yɨ̂p
3sg-pst.foc 3sg-buy-neg-buy nmlz-write 3sg-pst.foc 3sg-con.foc-steal
‘She didn’t BUY the book, she STOLE it.’
Similar to our analysis of negated reciprocals above, we take it that the suffix closest
to the verb attaches first as a result of verb raising. In (18), this is the negative suffix,
either -pɔ́ (18a) or -pé (18b). Negation thus precedes contrastive focus, and the negated
verb forms the input to the low focus position.
We propose (19) as the derivation of (18a):
9 Belletti
(2004) shows that, in Italian, low focus involving postverbal subjects is associated with new
information.
279
Philip T. Duncan, Travis Major & Mfon Udoinyang
(19)
TP
T’
T0
ú-ké
2sg-pst.foc
LFocP
ú- [ fɔ́pj -pɔ́ k ] -fɔ́p
2sg-burn-neg-burn
NegP
tj + tk
VP
tj
In the derivation of negative contrastive verb focus, LFoc0 probes for V0 (as in (16)
above), but it attracts the morphologically complex verb that has first raised to Neg0 .
The negative suffix is a consequence of V-to-Neg (similar to patterning of reciprocals),
and the negative suffix + a full verb copy are a consequence of V-to-Neg-to-Foc.
Why is affirmative contrastive focus a heavy CVV “prefix” while negative contrastive
focus retains a full copy? We tentatively propose (but leave for future analysis) the possibility that the grammar disprefers adjacent copies in contrastive focus constructions
and instead prefers to dissimilate and maintain distinction. Support for this comes from
other instances of contrastive focus in the language. Full reduplication exists elsewhere
in Ibibio, as in (20) below, but when items are contrastively focused some strategy for
differentiation is employed, as in (21):
(20)
a. ìtɔ́k
‘(a) race’
b. ìtɔ́k ìtɔ́k
‘hurriedly’
(21)
a. éwá ámì
‘this dog’
b. éwá ókò
‘that (visible) dog’
c. éwá ódò
‘that (not visible) dog’
280
a’. éwá ámì-ŋím̀mí
‘THIS dog (not that one)’
b’. éwá ókó-ŋóŋ̀kó
‘THAT (visible) dog (not this
one)’
c’. éwá ódò-ŋóǹdó
‘THAT (not visible) dog (not
this one)’
17 Searching high and low for focus in Ibibio
Thus, (phonologically) maintaining a distinction seems to be specific to contrastively
focused items – either verbs or demonstratives – in Ibibio; identical adjacent items in
non-contrastive constructions are permitted (20). The patterning of contrastively focused
demonstratives in (21) could be explained in a way that is analogous to the narrative of
contrastive verb focus we develop here; that is, if the syntax generates adjacent items
that are phonologically identical then the phonological system resorts to a post-syntactic
strategy to differentiate them.
In our analysis, then, verb focus morphology is a syntactic consequence of focused
verbs undergoing V-to-LFoc movement. This enables us to provide a more unified account of both affirmative and negative contrastive verb forms, since the same derivation
underlies both, despite their superficial dissimilarity. However, more work is needed
in this area to determine what additional morphophonological processes generate the
affirmative forms (such as those proposed by Akinlabi & Urua 2000; 2002). What we
see as critically important is that the presence of intervening material (e.g. the negative
suffix) blocks phonological reduction, though full copies of the verbs are present in the
syntactic derivation of both affirmative and negative forms.
5 Exhaustive focus
A second type of focus construction in Ibibio corresponds to exhaustive focus, which is
illustrated below in (22). As with argument focus – and unlike contrastive verb focus –
“focus” ké surfaces obligatorily in exhaustive focus constructions.
(22)
a. ìmá á-mà á-fèɰé ítɔ̀k
(input to 22b,c)
ima 3sg-pst 3sg-run race
‘Ima ran the race.’
Subject exhaustive focus
á-fèɰé ítɔ̀k
b. ìmá kpɔ́t á-ké
ima only 3sg-pst.foc 3sg-run race
‘Only Ima ran the race (not Ekpe or Akun).’/*‘Ima only ran the race (she
didn’t go to the party).’
c. ítɔ̀k kpɔ́t ké ìmá á-ké
á-fèɰé
Object exhaustive focus
race only comp ima 3sg-pst.foc 3sg-run
‘It was only the race that Ima ran.’
á-mà á-kót
àkàrà
d. èté â-ké-dép-pé
man 3sg-pst.foc-buy-rel bean.cake 3sg-pst 3sg-read
ŋ̀-wèt
(input to 22e)
nmlz-write
‘The man who bought the bean cake read the book.’
e. èté â-ké-dép-pé
àkàrà
kpɔ́t á-ké
á-kót
ŋ̀wèt
man 3sg-pst.foc-buy-rel bean.cake only 3sg-pst.foc 3sg-read nmlz-write
‘Only the man who bought the bean cake read the book (not Ima or
281
Philip T. Duncan, Travis Major & Mfon Udoinyang
Akun).’/*‘The man who bought the bean cake only read the book (he didn’t
read the magazine/he didn’t sell the book).’10
The focus particle kpɔ́t ‘only’ acts as an exhaustive focus operator, and it appears to the
right of the focused element. We posit that kpɔ́t heads its own phrasal projection, which
is a high focus phrase in the complementizer domain.11,12 Exhaustively focused XPs that
are attracted by HFoc0 thus land in Spec, HFocP (Rizzi 1997; Kayne 1998; É. Kiss 1998),
which guarantees that kpɔ́t always follows it’s focused constituent, as the examples in
(22) show. The structure in (23) shows the derivation of (22b) along these lines.
(23)
HFocP
DPsubj
ìmá
Ima
HFoc’
HFoc0
kpɔ́t
only
TP
DPsubj
T
á-ké
3sg-pst.foc
T’
VP
DPsubj
V’
V
DPOBJ
á-fèɰé
3sg-run
ítɔ̀k
race
10 An anonymous reviewer points out that this structure is ambiguous, and that it has the additional meaning
‘The man who bought only the bean cake read the book.’ We assume here that relative clauses of the form
in (22d-e) involve raising-to-C, which accounts for the appearance of the relative suffix. We take it that
this additional meaning is still compatible with movement to a high focus position, since the subject and
relativized verb also undergo movement to the C domain. However, we leave a more precise account of
relative clause structures for future investigation.
11
Note, too, that the complementizer ké is required when an object is exhaustively focused, as in (22c), and
that this complementizer appears after kpɔ́t. An overt C0 is illicit when subjects are exhaustively focused,
which is reminiscent of the subject-object asymmetry observed in argument focus constructions due to the
“that-trace effect” (see §3). It may be the case that kpɔ́t constructions do not require the type of Spec-Head
configuration that we propose. However, what is most important for our analysis is that exhaustive focus
in Ibibio recruits structure high in the left periphery.
12 An alternative analysis could treat kpɔt as a focus-sensitive adjunct much like ‘only’ in English (this point
́
was raised by an audience member of LSA 2015 and an anonymous reviewer). In such an account, kpɔ́t
would not head a projection in the C domain; instead, it would adjoin to an XP that bears a focus feature.
We take it that this is indeed a viable option, but at present it is difficult to distinguish with much certainty.
Importantly, data suggest that kpɔ́t -focused constituents (including exhaustive verb focus) activate a leftperipheral focus projection in a way that parallels argument focus and wh-questions in the language. That
is, exhaustive focus requires “focus” tense/aspect morphology (just like in cases of Ā-extraction), which is
not a requirement of verb focus constructions that recruit the low focus projection.
282
17 Searching high and low for focus in Ibibio
When verbs are exhaustively focused in Ibibio, the entire TP is targeted for movement.
Consequently, the exhaustive focus operator kpɔ́t always appears to the right edge of the
TP, as seen in (24).13
(24)
a. [HFocP [TP ìmá á-ké
á-fèɰè ítɔ̀k ] kpɔ́t ]
ima 3sg-pst.foc 3sg-run race only
‘Ima only ran the race.’/*‘Only Ima ran the race.’
b. [HFocP [TP ékpê á-ké
á-kòt
ŋ̀-wèt
] kpɔ́t ]
ekpe 3sg-pst.foc 3sg-read nmlz-write only
‘Ekpe only read the book.’/*‘Only Ekpe read the book.’
Moreover, exhaustive verb focus constructions bear an affinity to subject focus in that
an overt complementizer is not permitted:
(25)
á-fèɰé ítɔ̀k kpɔ́t ké
* ìmá á-ké
ima 3sg-pst.foc 3sg-run race only comp
(Intended: ‘Ima only ran the race.’)
Exhaustively focused verbs thus demonstrate a “that-trace effect,” which is the same
configuration for subject focus.
We propose the analysis for exhaustive TP focus shown in (26), where HFoc0 probes
for V0 and pied-pipes (Ross 1967) TP.
(26)
HFocP
TPi
HFoc’
HFoc0
ti
kpɔ́t
Our analysis of (24a) is given in (27):
13
One anonymous reviewer notes that, with a change of tone, the second readings in (24) are possible. Thus, if
the final tone of the verb complexes in (24a) and (24b) are high, the grammaticality judgments are reversed.
We assume that the different readings have similar underlying structures. The fact that kpɔ́t can scope over
the whole TP or the subject, but not the object, further supports our pied-piping analysis. It could be that
the object is too deeply embedded inside TP to be focused in this construction.
283
Philip T. Duncan, Travis Major & Mfon Udoinyang
(27)
HFocP
TPi
HFoc’
T’
ìmá
Ima
á-ké
3sg-pst.foc
kpɔ́t
only
ti
VP
á-fèɰé
3sg-run
ítɔ̀k
race
Exhaustive verb focus constructions thus require a different structural configuration
than that of contrastively focused verbs. Exhaustive VP focus targets XPs for movement
to the C layer, rather than being derived by head movement more local to VP. Exhaustively focused constituents move to Spec,HFocP and appear to the left of the exhaustive
focus operator.
6 Double focus
Given the tree in (1), Ibibio should allow “real multiple focus” (Krifka 1992). That is, if
there are truly two distinct focus projections, it should be able to activate both focus
heads in a single clause; Ibibio shows that this can indeed happen.
Rizzi (1997: 298) noted for Italian that wh-questions are incompatible with focus constructions, since they compete for the same position:
(28)
Italian
a. * A chi IL PREMIO NOBEL dovrebbero dare?
‘To whom THE NOBEL PRIZE should they give?’ (Rizzi 1997: 298)
b. * IL PREMIO NOBEL a chi dovrebbeo dare?
‘THE NOBEL PRIZE to whom should they give?’ (Rizzi 1997: 298)
In Ibibio, wh-questions allow movement to the left periphery, while contrastive verb
focus is derived in the inflectional domain. As a result, Ibibio contrastive verb focus is
compatible with wh-questions, as seen in (29) below.
(29)
284
a. ǹsǒ ké (àfò) à-dìá-díá
what comp 2sg 2sg-con.foc-eat
‘What the hell are you EATING?’
b. ǹsǒ ké (àfò) mmé-ú-ké-ú-dìá-ɰá-díá
what comp 2sg mmé-2sg-pst.foc-2sg-con.foc-neg-eat
‘What the hell didn’t you EAT?’
17 Searching high and low for focus in Ibibio
In (29), the left edge HFoc and lower LFoc are both activated, allowing the wh-element
ǹsǒ ‘what?’ to move to the C domain and the verb dìá ‘eat’ to raise to the lower focus position. This type of double focus interestingly produces a wh-the-hell reading (Pesetsky
1987; den Dikken & Giannakidou 2002).
Ibibio also permits double focus (contrastive + exhaustive) with verbs probed for by
both Foc heads:
(30) é-ké
é-bɔ̀ɔ́-bwɔ́t
ákʌ́k kpɔ́t
3pl-pst.foc 3pl-con.foc-borrow money only
‘They only BORROWED money.’ (Response to: ‘Did they steal money?’)
As with (29), our analysis can account for the simultaneity of these focus types since
they correspond to distinct structural configurations.
We propose the (truncated) structure in (31) for double verb focus in Ibibio:
(31)
Double verb focus
HFocP
HFoc’
TP
kpɔ́t
T0
LFocP
LFoc0
VP
V0
DP
Here, the order in the derivation is critical, and the derivation of contrastive verb
focus precedes exhaustive verb focus. Accordingly, LFoc0 probes for V0 and causes it to
raise, which ensures the verb focus morphology unique to contrastively focused verbs.
Following this, HFoc0 also probes for V0 and pied-pipes the whole TP to Spec,HFocP
since exhaustive focus constructions require phrasal movement. This ensures that the
contrastively focused verb along with the object DP surface to the left of the exhaustive
focus operator. Our analysis of (30) is seen below in (32):
(32)
HFocP
[TP é-ké [LFoc [é-bɔ̀ɔ́-bwót àkʌ́k ]]i
3pl-pst.foc 3pl-con.foc-borrow money
HFoc’
kpɔ́t
only
ti
285
Philip T. Duncan, Travis Major & Mfon Udoinyang
Again, in our proposal contrastively focused verbs must be derived prior to piedpiping of the TP for verb focus morphology to occur on the verb in a double verb focus
construction.14
To summarize, in this section we presented data to show that Ibibio does allow double
focus constructions, and that such constructions involve real multiple focus. Given the
distinct structural configurations required for exhaustive and contrastive verb focus, our
analysis comports rather nicely with these facts. Further, given that the different focus
positions correspond to particular semantic interpretations when verbs are focused, our
proposal also accounts for the scope effects present in double verb focus.
7 Conclusions
We have argued in this paper that Ibibio verb focus constructions are not unified. We motivated the existence of two types of syntactically-driven focus constructions involving
verbs in Ibibio: verb raising to a low focus position in the inflectional domain corresponds
to contrastive focus, and TP pied-piping to the C layer corresponds to exhaustive focus.
Since these focus types are structurally distinct, both Foc probes can target V. Thus, double verb focus is permitted in Ibibio (Krifka 1992), and exhaustively focused verbs always
take wide scope (Krifka 1992; Kiss 1998) over contrastively focused ones. From a typological perspective, Ibibio verb focus constructions are significant in that they provide
independent evidence for a low focus position associated with a “specialized” semantic
interpretation (Belletti 2004). The low focus position in Ibibio is rather unique, however,
in that it seems to exclusively target verbs and not, say, NPs. Further exploration into
Ibibio contrastive verb focus could thus yield interesting theoretical insights into the
nature of low focus and what is possible cross-linguistically.
14
Our analysis predicts that exhaustive focus always takes wide scope over contrastive focus, which is indeed
borne out:
(i)
ŋ̀-wèt
] kpɔ́t ]
á-kót
a. ìyó, ékpê [HFocP [TP á-ke
3sg-pst.foc 3sg-read nmlz-write only
no ekpe
‘No, Ekpe only [read the book] (not the magazine/he didn’t even do his laundry/*he did not
take it away).’
b. ìyó, ékpê [HFocP [LFocP á-ke
á-kòó-kót
ŋ̀-wèt
] kpɔ́ t ]
no ekpe
3sg-pst.foc 3sg-con.foc-read nmlz-write only
‘No, Ekpe only [READ the book] (he did not take it away/*he didn’t even do his laundry/*not
the magazine).’
Thus, (i.b) only corresponds to the interpretation where reading (not doing something else to) the book is
the only thing that Ekpe did; it cannot mean that some object other than the book was read.
286
17 Searching high and low for focus in Ibibio
Abbreviations
1
2
3
comp
con
cond
foc
impf
indf
1st person
2nd person
3rd person
complementizer
contrastive
conditional
focus
imperfective
indefinite
neg
nmlz
obj
perf
pl
pst
sg
S/subj
negation
nominalizer
object
perfective
plural
past
singular
subject
Ibibio is tonal, and tones are marked in the following manner:
V́
V̀
V̂
high tone
low tone
falling tone (note that tones are marked on either vowels or syllabic nasals in the
data).
Acknowledgements
We would like to thank our fellow participants in the 2014 Field Methods in Linguistic
Description class and the 2014 Research in Field Linguistics seminar at KU, and audience members at both ACAL 45 and LSA 2015 for helpful comments and discussions.
We are especially grateful to Harold Torrence and Jason Kandybowicz for their many
insights and critiques. Edopeseabasi Udoinyang also graciously provided examples and
judgments that informed the final version of this paper. Finally, we thank two anonymous reviewers for their help in suggesting improvements. Unless otherwise noted, the
data used in this paper are from Mfon and Edopeseabasi and reflect their judgments.
References
Akinlabi, Akinbiyi & Eno E. Urua. 2000. Tone in ibibio verbal reduplication. In H. Ekkehard Wolff & Orin D. Gensler (eds.), Proceedings of the 2nd World Congress of African
Linguistics Leipzig 1997, 280–291. Köln: Rüdiger Köppe.
Akinlabi, Akinbiyi & Eno E. Urua. 2002. Foot structure in the Ibibio verb. JALL 23. 119–
160.
Baker, Mark C. 1985. The Mirror Principle and morphosyntactic explanation. Linguistic
Inquiry 16(3). 373–415.
Baker, Mark C. 2008. On the nature of the antiagreement effect: Evidence from wh-insitu in Ibibio. Linguistic Inquiry 39. 615–632.
Baker, Mark C. & Willie U. Willie. 2010. Agreement in Ibibio: From every head, to every
head. Syntax 13(2). 99–132.
287
Philip T. Duncan, Travis Major & Mfon Udoinyang
Belletti, Adriana. 2004. Aspects of the low IP area. In Luigi Rizzi (ed.), The structure of CP
and IP, 16–51. Oxford: Oxford University Press.
Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Roger Martin, David
Michaels & Juan Uriagereka (eds.), Step by step: Essays on minimalist syntax in honor
of Howard Lasnik, 89–156. Cambridge, MA: MIT Press.
Chomsky, Noam. 2001. Derivation by phase. In Michael Kenstowicz (ed.), Ken Hale: A
life in language, 1–52. Cambridge, MA: MIT Press.
Chomsky, Noam & Howard Lasnik. 1977. Filters and control. Linguistic Inquiry 8. 425–
504.
Cook, Thomas L. 2002. Focus in Efik. JALL 23. 113–152.
den Dikken, Marcel & Anastasia Giannakidou. 2002. From hell to polarity: “Aggressively
non-D-linked” wh-phrases as polarity items. Linguistic Inquiry 33(1). 31–61.
Essien, Okon E. 1990a. A grammar of the Ibibio language. Ibadan: University Press Limited.
Essien, Okon E. 1990b. The aspectual character of a verb and tense in Ibibio. Journal of
West African Languages 20. 64–72.
Holmberg, Anders & Christer Platzack. 1995. The role of inflection in Scandinavian syntax.
Oxford: Oxford University Press.
Kayne, Richard S. 1998. Overt vs. Covert movement. Syntax 1(2). 128–191.
Kiss, Katalin É. 1998. Identificational focus versus information focus. Language 74(2).
245–273.
Krifka, Manfred. 1992. A compositional semantics for multiple focus constructions. In
Joachim Jacobs (ed.), Informationsstruktur und grammatik, 17–53. VS Verlag: Opladen.
Perlmutter, David M. 1971. Deep and surface structure constraints in syntax. New York, NY:
Holt, Rinehart & Winston.
Pesetsky, David. 1987. Wh-in-situ: Movement and unselective binding. In Eric Reuland &
Alice G. B. ter Meulen (eds.), The representation of (in)definiteness, 98–130. Cambridge,
MA: MIT.
Pollock, Jean-Yves. 1989. Verb movement, Universal Grammar, and the structure of IP.
Linguistic Inquiry 20(3). 365–424.
Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of grammar: A handbook in generative syntax, 281–337. Kluwer: Dordrecht.
Ross, John R. 1967. Constraints on variables in syntax. Cambridge, MA: MIT Doctoral
dissertation.
Willie, Willie U. & Mfon. E Udoinyang. 2012. The missing component in Ibibio linguistic
time. In Michael R. Marlo, Nikki B. Adams, Christopher R. Green, Michelle Morrison &
Tristan M. Purvis (eds.), Selected proceedings of the 42nd Annual Conference on African
Linguistics, 240–256. Somerville, MA: Cascadilla.
288
Chapter 18
More on have and need
Claire Halpert
University of Minnesota
Michael Diercks
Pomona College
This paper addresses recent work on the cross-linguistic patterns involving have and need
predicates, focusing on the debate surrounding the claim that all languages that lack a transitive have also lack transitive need (Harves & Kayne 2012). In this paper, we move the
discussion beyond these surface patterns, first by presenting new syntactic diagnostics to
demonstrate that the Bantu language counter-examples to the proposed generalization discussed by Antonov & Jacques (2014) are true counter-examples to the original claim by
Harves & Kayne (2012). From this perspective, we evaluate the relevance of these conclusions for Harves & Kayne (2012)’s lexical decomposition analysis of need. We conclude that
although Bantu languages form a straightforward counter-example to the proposed Harves
& Kayne (2012) typology, as Antonov & Jacques (2014) noted, there are in fact some deep
similarities between the Bantu patterns and the proposals of Harves & Kayne (2012). In
particular, we suggest that their observations about the role of case in the distribution of
have and need verbs may in fact be amenable to the Bantu patterns, suggesting that their
conclusions cannot yet be abandoned.
1 Overview of the Issues
Harves & Kayne (2012) survey a number of languages and propose an empirical generalization: all languages that lack a lexical verb of possession (have) likewise lack a transitive lexical verb need. Based on this apparent typological gap they propose a lexical
decomposition analysis of need. In response, Antonov & Jacques (2014) provide a range
of typological data showing that the typological generalization that Harves & Kayne
(2012) rely on is not in fact surface-true, a conclusion this paper supports. The chart in
Table 1 summarizes both Harves & Kayne (2012)’s original typological conclusions and
the additions of Antonov & Jacques (2014) and this paper, listed in bold.
Claire Halpert & Michael Diercks. More on have and need. In Jason Kandybowicz, Travis Major,
Harold Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected papers
from the 45th Annual Conference on African Linguistics, 289–305. Berlin: Language Science Press.
DOI:10.5281/zenodo.1251742
Claire Halpert & Michael Diercks
Table 1: Revised typology of possession and need, with additions in bold
Languages with
transitive ‘need’
Languages without
transitive
‘need’
H-languages
B-languages
Czech, Slovak, Polish,
Slovenian, Croatian, Servian
(dialects), Belorussian,
English, German, Yiddish,
Luxemburgish, Dutch,
Swedish, Norwegian,
Icelandic, Spanish, Catalan,
Basque, Paraguayan Guaraní,
Purépecha (Tarascan),
Mapudungun
Zulu, Setswana, Kuria,
Swahili, Otjiherero, Estonian,
Moroccan Arabic, Algerian
Arabic, Likpe, Ewe,
Ayacucho Quechua
Bulgarian, Serbian (standard),
Lithuanian, French, Italian,
Bellinzonese, Portuguese,
Romanian, Farsi, Armenian,
Albanian, Latin, Ancient
Greek
Russian, Latvian, Sakha,
Bhojpuri, Bengali, Hindi,
Marathi, Irish, Welsh, Scots
Gaelic, Georgian, Hungarian,
Turkish, Korean, Peruvian
Quechua (Cuzco, Cajamarca,
Huallaga), Bolivian Quechua,
Yucatec Maya, Tamil,
Mohawk, Amharic
1.1 An empirical correction to Harves & Kayne (2012)
Harves & Kayne (2012) focus on what they claim is a significant typological gap in the
cross-linguistic expression of possession and need, formalized in (1):
(1)
Harves-Kayne Generalization (Strong version):
(Harves & Kayne 2012: 1)
All languages that have a transitive verb corresponding to need are H-languages.
The gap in their data occurs when we compare languages that use a transitive verb
of possession, or H(ave)-languages, to languages that use a non-transitive strategy to
express possession, or B(e)-languages. While possession in H-languages looks straightforwardly transitive, involving a nominative-accusative case pattern, possession in Blanguages does not: in B-language possessors are typically oblique, and possessees are
nominative instead of accusative (unlike possessees in H-languages). H-languages may
or may not have a transitive need verb, but Harves & Kayne (2012) crucially claim that
B-languages never do.
290
18 More on have and need
(2) H-language with transitive need: Czech
a. Mají
nové auto.
have.3pl new car.acc
‘They have a new car.’
b. Tvoje děti
tě
potřebují.
your children.nom you.acc need.3pl
‘Your children need you.’
(3)
(Harves & Kayne 2012: 4a, 5a)
H-language with non-transitive need: French
(Harves & Kayne 2012: 6a, 7a)
a. J’ ai
une voiture.
I have.1sg a car
‘I have a car.’
b. J’ ai
besoin d’ une voiture.
I have.1sg need of a car
‘I need a car.’
(4) B-language with non-transitive possession: Latvian (Harves & Kayne 2012: 2b,
3c)
a. Man ir velosipēds.
me.dat is bicycle.nom
‘I have a bicycle.’
b. Man vajag
dakšu.
me.dat need.3sg fork.gen
‘I need a fork.’
Harves & Kayne (2012) argue that this crucial gap – the absence of B-languages with
transitive need – follows directly from an incorporation account of transitive need: the
derivation of the verb need involves incorporation of a nominal ‘need’ into an unpronounced (transitive, abstract) HAVE. Because ‘need’ incorporates, it does not require
case (Baker 1988), which allows HAVE to assign accusative to the object. Languages that
lack an overt have verb are assumed to lack abstract HAVE and are thus unable to do the
necessary incorporation to create transitive need.
(5)
VP
N+V
[need𝑖 + HAVE]
NP
t𝑖
DP
291
Claire Halpert & Michael Diercks
As noted by Antonov & Jacques (2014), a common pattern in Bantu languages contradicts the generalization in (1): the following examples show 3 languages that have
transitive lexical verbs for need but construct predicative possession using a copular (be)
construction followed by the preposition with.1
(6) Zulu: be-possession and transitive need
a. ngi- zo- ba nemali.
1sg- fut- be with.aug- 9money
‘I will have money.’
b. ngi- zo- dinga imali.
1sg- fut- need aug.9money
‘I will need money.’
(7)
Swahili: be-possession and transitive need
a. ni- li- kuwa na nyumba.
1sg- pst- be
with 9house
‘I had a house.’
b. ni- li- hitaji nyumba.
1sg- pst- need 9house
‘I needed a house.’
(8)
Kuria: be-possession and transitive need
a. Gati n- a- are neng’ɔɔmbe.
1Gati foc- 1s- rem.pst- be with- 9cow
‘Gati had a cow.’ (remote past)
b. Gati n- a- atun- ire
eng’ɔɔmbe.
1Gati foc- 1s- rem.pst- need- rem.pst 9cow
‘Gati needed a cow.’ (remote past)
In addition to the languages shown here, initial evidence suggests that this pattern
is well attested throughout the Bantu family. Herero, for example, expresses predicative
possession using a be (with) construction, na, but has a transitive verb of need, hepa,
that is distinct from the verb of wanting vanga (Nguako 2013). Setswana also uses a be
(with) construction, na (le), for predicative possession;2 transitive tlhoka for ‘need’; and
batla for ‘want’. We include these languages in table 1 on the basis of this preliminary
evidence. Other languages, including Shona, Lubukusu, and Tiriki also lack a transitive
verb have and express ‘need’ with a lexical verb; these differ, however, in that they seem
to collapse need and want (relying on circumlocutions in sentences contrasting ‘desiring’
1
2
Antonov & Jacques (2014) give parallel data to ours here in Swahili and Zulu.
Creissels (2013) observes, though, that in Setswana predicative possession patterns in some respects like
a transitive verb. He remarks that this pattern is a departure from the general Bantu pattern in which
predicative possession is completely indistinguishable from the comitative construction.
292
18 More on have and need
with ‘needing’).3 No Bantu language that we examined expressed predicative possession
via a transitive have verb.
It is clear from these Bantu examples that the generalization in (1) is not surface-true:
these languages all have lexical verbs for need but lack a lexical verb have. Antonov
& Jacques (2014) make this argument based on data like these from Swahili and Zulu,
as well as similar data from a typologically diverse set of languages (including Arabic,
Quechua, and Kwa languages). Our departure point is to investigate the issue in more
syntactic depth to determine whether these apparent counter-examples hold up under
further investigation and, if so, what the consequences are for the Harves & Kayne (2012)
analysis of need.
We suggest in this paper that the resulting picture is more nuanced. While the Swahili
and Zulu patterns indeed constitute true counter-examples to Harves & Kayne (2012)’s
generalization in (1), Harves & Kayne (2012)’s revised generalization, discussed below in
(10), and a more in-depth consideration of structural licensing suggests that their core
intuitions may still have merit, at least with respect to the Bantu data. This conclusion
contrasts with that of Antonov & Jacques (2014), who state on the basis of the typological
evidence that Harves & Kayne (2012)’s “hypothesis is thus unlikely to be valid as an
absolute universal.” While their conclusion may ultimately be correct, we suggest that
a revised conception of Harves & Kayne (2012)’s relevant generalization based on the
Bantu evidence could potentially reveal a modified universal structural decomposition of
need verbs. This proposal makes useful predictions about the structure of these predicates
in the other languages in Antonov & Jacques (2014)’s study, setting the stage for future
investigation.
While Antonov & Jacques (2014) establish a number of potential counter-examples
to Harves & Kayne (2012)’s proposed typology,4 Harves & Kayne (2012) themselves discuss one language that does not straightforwardly follow their generalization: Finnish is
canonically described as a B-language but nonetheless has a transitive need verb with a
nom-acc case pattern. Harves & Kayne (2012) point out that while Finnish uses the same
be verb in existential, locative, predicational, and possessive sentences, possessives differ from the other constructions in taking an accusative – rather than a nominative –
object:
3
Kuria, illustrated in (8) and to which we do not return in this paper, appears at a glance to fall into this
category: our consultant reports that the verb ugu-tuna ‘inf-need’ in Kuria can also have a reading of
‘want.’ Despite this apparent lexical overlap, it is possible to contrast ugu-tuna with an unambiguous verb
of desire, uku-igomba, producing a sentence like Gati naigombere imburi, si bono natunire en’gombe ‘Gati
desired a goat, but he needed a cow.’ This kind of sentence would be unlikely if ugu-tuna was lexically a
verb of ‘wanting’ just as much as ‘needing’ (cf. English #John desired a Porsche, but wanted a family sedan.).
We suspect, therefore, that Kuria’s ugu-tuna verb is probably best classified as a true ‘need’ verb, with
metaphorical extensions to notions of ‘wanting’ (cf. English I need a beer right now). On this basis we
include Kuria in the languages added in Table 1. Due to this complication, however, we restrict the core
examples discussed in the paper to Swahili and Zulu.
4
See Kayne (2014), though, for a discussion of some problems with the evidence Antonov & Jacques (2014)
give.
293
Claire Halpert & Michael Diercks
(9) Finnish predicational vs. possessive be
a. Hän
on
vanha.
he.nom be.3sg old.nom
‘He is old.’
b. Minu- lla
on
häne-t.
Iadess be.3sg him-acc
‘I have him.’
(Harves & Kayne 2012: 14c, 13)
Harves & Kayne (2012) argue that the accusative case assignment in (9) crucially distinguishes Finnish from other B-languages: because Finnish expresses possession via an
accusative-assigning (B)-verb, need may incorporate into accusative-assigning BE in this
language to yield the transitive need pattern. They thus revise their generalization to reflect the importance of case-assignment patterns, as opposed to BE/HAVE distinctions:
(10) Harves-Kayne Generalization (revised):
(Harves & Kayne 2012: 15)
All languages that have a transitive verb corresponding to need are languages that
have an accusative-case-assigning verb of possession.
As we will see in the next section, even the revised generalization does not seem to
capture the Bantu facts: the possessee in Swahili and Zulu possessive predication does
not behave like a normal transitive direct object, but instead exhibits similar behavior
to the ‘objects’ of copular, existential, and locative predication, which also involve be.
In §3, we return to an aspect of the generalization in (10) without a clear connection to
the Bantu data – case assignment patterns. We propose that the the Bantu exceptions
to Harves & Kayne (2012)’s generalization(s) may in fact be linked to the exceptional
behavior of Bantu languages with respect to syntactic case. Based on recent work on
case in Bantu (e.g. Diercks 2012; Halpert 2012), we suggest that case-licensing of objects is
independent of transitivity in these languages; transitive verbs and B-constructions have
identical licensing properties. Given this pattern, a version of (10) that simply requires
identical licensing properties between predicative possession and transitive verbs may
be tenable.
2 The Bantu examples are true counter-examples
We have already established that the surface generalization in (1) cannot be upheld in
the face of the patterns in Swahili and Zulu,5 which are both B-languages that nonetheless have a lexical verb need. In this section, we demonstrate that possessees in these
languages are not canonical transitive objects, which rules out a Finnish-style analysis
for the Bantu facts.
As is common in the Bantu family, neither Zulu nor Swahili has overt case morphology, instead marking most grammatical relations on the verb itself via subject marking
5 As
well as the other Bantu languages discussed above. In addition, while we focus on Zulu and Swahili
here, we note that Kuria exhibits identical behavior on all relevant diagnostics.
294
18 More on have and need
and object marking. This lack of case morphology means that we cannot simply use
nominal morphology to evaluate Harves & Kayne (2012)’s generalizations. Instead, we
focus on object marking and A-bar extraction as tests for transitive-object behavior. As
the following patterns demonstrate, possessees in Zulu and Swahili show distinct properties from canonical transitive objects, suggesting that they are true counter-examples
to the generalizations in (1)/(10) and not instances of covert canonical objects.
2.1 Object markers for need and have
Most Bantu languages can mark transitive objects on the verb via a morpheme that
precedes the stem and follows other inflectional material (see Riedel 2009; Marten et al.
2007; Zeller 2012; Bax & Diercks 2012, a.o., for additional discussion). Abstracting away
from particular analyses of object markers, we instead take their availability to be a
canonical property of transitive objects. As Swahili shows in (11), need uses the normal
pre-stem OM to pronominalize an object, just as the transitive verb want does, while
predicative possession requires an exceptional enclitic morpheme to pronominalize an
object. A pre-stem object marker (11d) is ungrammatical.
(11) Swahili
a. Gati a- li- i- taka.
1Gati 1s- pst- 9o- want
‘Gati wanted it (a house).’ (remote past)
b. Gati a- li- i- hitaji.
1Gati 1s- pst- 9o- need
‘Gati needed it (a house).’ (remote past)
c. Gati a- li- kuwa na- yo.
1Gati 1s- pst- be
with- 9pronoun
‘Gati had it (a house).’ (remote past)
d. * Gati a- li- i- kuwa na- (yo)
1Gati 1s- pst- 9o- be
with- 9pronoun
The examples (12) illustrate the same pattern for Zulu:
(12) Zulu
a. ngi- zo- yi- funa.
1sg- fut- 9o- want
‘I will want it (money).’
b. ngi- zo- yi- dinga.
1sg- fut- 9o- need
‘I will need it (money).’
c. ngi- zo- ba na- yo.
1sg- fut- be with- 9pronoun
‘I will have it (money).’
295
Claire Halpert & Michael Diercks
d. * ngi- zo- yi- ba na- (yo).
1sg- fut- 9o- be with (9pronoun)
These contrasts show that the canonical object marking patterns that are available
for objects of transitive verbs are not available for possessees in predicative possession
for Swahili and Zulu (see Antonov & Jacques 2014 for similar discussion). This pattern
suggests that the Swahili and Zulu counter-examples are not instances of a transitive-like
construction in disguise.
2.2 Object extraction for need and have
Extraction patterns provide an additional argument for distinguishing between possessee
arguments and transitive objects. In Swahili, for example, the verb need shows the same
patterns for object extraction as transitive verbs: an object operator can simply be A’moved to the left periphery. In contrast, such dislocation in predicative possession requires a resumptive clitic:
(13)
Swahili object extraction
a. Ni- li- ona ki- tabu amba- cho Gati a- li- nunua.
1sgs- pst- see 7- book comp- 7rel 1Gati 1s- pst- buy
‘I saw the book that Gati bought.’
b. Ni- li- ona ki- tabu amba- cho Gati a- li- kuwa na- cho.
with- 7pro.
1sgs- pst- see 7- book comp- 7rel 1Gati 1s- pst- be
‘I saw the book that Gati had.’
The requirement of a resumptive enclitic here is exceptional among instances of object
extraction in Swahili. Notably, it is not exceptional for predicative possession in other
Bantu languages. Zulu again shows the same patterns, distinguishing transitive object
extraction, which requires object marking, from extraction of a possessee, which requires
the enclitic:
(14)
296
Zulu object extraction
a. y- imalini e- ngi- zo- yi- funa?
cop- aug.9money- what rel- 1sg- fut- 9o- want
‘How much money will I want?’
b. y- imalini e- ngi- zo- yi- dinga?
cop- aug.9money- what rel- 1sg- fut- 9o- need
‘How much money will I need?’
c. y- imalini e- ngi- zo- ba na- yo?
cop- aug.9money- what rel- 1sg- fut- be with- 9pronoun
‘How much money will I have?’
d. * y- imalini e- ngi- zo- yi- ba na- (yo)?
cop- aug.9money- what rel- 1sg- fut- 9o- be with- (9pronoun)
18 More on have and need
If Swahili and Zulu possessive constructions were transitive verbs disguised as Bconstructions, we would expect parallel behavior between transitives and possessives,
contrary to fact.
In fact, a closer parallel to the extraction properties of possessive constructions is extraction from a prepositional phrase, which also requires a resumptive enclitic pronoun:
(15)
Zulu PP extraction
a. ubhuthi
e- ngi- hamba na- ye
(u)- ng- uSipho.
aug.1brother rel- 1sg- go
with- 1pronoun (1s)- cop- aug.1Sipho
‘The guy I’m going with is Sipho.’
b. ubhuthi e-ngi-khuluma nga- ye u- zo ba (ng)- umongameli.
aug.1br rel-1sg-speak instr- 1pro 1s- fut- be (cop)- aug.1president
‘The guy who I’m talking about will be president.’
(16) Swahili PP extraction
a. mw- anafunzi ni- na- ye- enda na- ye
ni Gati.
1- student 1sg- pres- 9rel- go with- 1pronoun is 1Gati
‘The student who I’m going with is Gati.’
b. m- tu
ni- na- ye- zungumza na- ye
a- ta- kuwa
1- person 1sg- pres- 9rel- converse with- 1pronoun 1s- fut- be
rais.
9president
‘The person who I’m talking to will be president.’
In short, A’-extraction in predicative possession patterns with extraction of obliques
– and not with extraction of direct objects. This pattern is consistent with an analysis
of the possessive constructions in Swahili and Zulu as a copula plus a prepositional
phrase, exactly what it appears to be on the surface. This evidence therefore supports the
conclusion that Swahili and Zulu are true B-languages (expressing possession via a basic
copular construction), and therefore true counter-examples to the (1)/(10) generalization.
2.3 Predicative possession as a non-verbal construction
An additional parallel between predicative possession and other copular constructions
in Bantu is found in the distribution of the be verb. The examples we have seen involve
a be verb plus the preposition (some version of na or ne in all the languages considered
here). More generally, the full verbal form appears only as needed to host overt tense
morphology; in present tense constructions, for example, we find a reduced structure
with only agreement and the preposition in many languages, as Zulu and Swahili show:
(17)
a.
ngi- nemali.
1sg- with.aug- 9money
‘I have money.’
[Zulu]
297
Claire Halpert & Michael Diercks
b.
ni- na nyumba.
1sg- with 9house
‘I have a house.’
[Swahili]
This same pattern occurs in copular clauses, with the full copular verb (and inflection)
only appearing in non-present tenses, as the examples above show in (15) for Zulu and
in (16) for Swahili. Buell & de Dreu (2013) provide a detailed comparison of non-verbal
predication in Zulu, demonstrating that possessive predication exhibits parallel behavior
to other copular constructions, in line with what we have shown here.
2.4 Intermediate summary
What we have seen in this section is that, based on evidence from both object marking and object extraction, possessees in predicative possession constructions do not display canonical properties of transitive objects. Without overt accusative case-marking
in Bantu languages, these canonical object properties are the best means to examine
whether the revised generalization in (10) holds up in the face of the Bantu counterexamples. The kind of exceptional copula behavior of the be-possessive in Finnish does
not extend to Swahili and Zulu, which appear to be truly copular-based constructions.
This conclusion was bolstered by the observation that the be-verb in these contexts appears to pattern in normal ways for a copula, being null in the present tense. These facts
thus allow us to move beyond Antonov & Jacques (2014)’s observation that Bantu languages form a surface counter-example to Harves & Kayne (2012)’s generalization to
show that the counter-examples hold even on deeper syntactic measures of transitivity
and objecthood.
3 The role of case
In the previous section, we concluded that Bantu languages like Zulu and Swahili constitute a robust counter-example to Harves & Kayne (2012)’s generalization about have and
need. As we saw in (10), however, the revised version of their claim specifically refers to
the case assignment properties of the relevant predicates, with the idea that languages
where possessees receive acc have transitive need (that assigns acc to its direct object).
Given the Bantu counter-examples, we can draw one of two possible conclusions. First,
we might conclude that Harves & Kayne (2012)’s generalizations are empirically inaccurate and their resulting analysis of the decomposition of need is therefore untenable.
A second alternative would be that Harves & Kayne (2012)’s revised generalization in
(10) is on the right track, with the distribution of H- and B- languages relating to the the
presence of transitive need based on the availability of Case-licensing.
This second alternative is not transparently correct: the surface forms show no evidence that objects of predicative possession and need are Case-licensed identically in
Zulu and Swahili. We noted in the previous section that while Bantu languages show
298
18 More on have and need
no obvious morphological case marking on nominals, evidence from structural diagnostics demonstrates clear syntactic distinctions between the objects of transitive predicates
and possessees in predicative possession. In the following subsections we nonetheless
return to the issue of Case and what role its presence (or absence) might play in the
Bantu possession pattern.
3.1 Another test for have and need
As we saw above, need in Swahili and Zulu always patterns with transitive verbs – and
not with have (i.e. BE + P) constructions. This raises a critical question, however: are
have and need always syntactically different in Bantu? The short answer is maybe not:
one morphosyntactic pattern in Zulu, augment distribution, in fact suggests that both
types of object are licensed in the same way.
Zulu nouns are typically marked with an initial augment vowel that appears before
the noun class prefix. This augment vowel can be dropped on some indefinites6 in certain
syntactic positions – in particular, immediately after the verb inside vP (Halpert 2012).
We show in the data that follow that with respect to augment drop, have, need, and transitive verbs all behave in the same way in Zulu. As the data in (18) show, in the relevant
contexts (triggering NPI readings in these examples) an augment may be dropped on the
highest DP after a transitive verb. Unsurprisingly, need shows the same behavior in (19).
(18) Zulu: augment drop possible on highest DP after transitive verb
a. ngi- bona u-muntu.
1sg- see aug-1person
‘I see someone/the person.’
b. A- ngi- bon- i
muntu.
neg- 1sg- see- neg 1person
‘I don’t see anybody.’
(19) Zulu: augment drop possible on highest DP after need
a. ngi- dinga i-mali.
1sg- need aug-9money
‘I need money.’
b. A- ngi- ding- i
mali.
neg- 1sg- need- neg 9money
‘I don’t need any money.’
What distinguishes this test from those in the previous section is that, as (20) shows,
the possessee in predicative possession behaves like a transitive object, allowing the
augment to be dropped in the relevant syntactic contexts:
6
While nonveridical environments are typically necessary for augment drop, Halpert (2012) demonstrates
that there are additional, independent syntactic conditions under which the process is licensed, on which
we focus here.
299
Claire Halpert & Michael Diercks
(20) Zulu: augment drop possible on possessee with predicative possession na
a. ngi- ne-mali.
(na+imali)
1sg- with.aug-9money
‘I have money.’
b. A- ngi- na mali.
neg- 1sg- with 9money
‘I don’t have any money.’
The behavior of the possessee in (20) is not unique to the possessive constructure,
however; rather, it seems to be a property of the na preposition more generally that
augment drop is permitted on its complement under the right structural conditions:
(21)
Zulu: preposition na allows augment drop when highest element in vP
a. u-Mfundo
u- dlala i-bhola nomuntu. (na+umuntu)
aug-1Mfundo 1s- play aug-5ball na.aug- 1person
‘Mfundo is playing soccer with someone/the person.’
b. u-Mfundo
a- ka- dlal- i
na- muntu i-bhola.
aug-1Mfundo neg- 1s- play- neg na- 1person aug-5ball
‘Mfundo isn’t playing soccer with anyone.’
c. *u-Mfundo a- ka- dlal- i
i-bhola na- muntu.
aug-1Mfundo neg- 1s- play- neg aug-5ball na- 1person
Crucially, the na preposition contrasts with certain other prepositions in the language.
While na PPs are essentially transparent with respect to the constraints on augment drop,
some prepositions do not alternate, instead always requiring the no-augment version
regardless of position or interpretation, as shown for kwa- and ku- below:
(22) Zulu: prepositions kwa- and ku- prohibit augment on their complement
a. u-Sipho
u- zo- pheka ukudla
kwa- zingane/ *kwezingane.
aug-1Sipho 1s- fut- cook aug.15food kwa- 10child/ *kwa.aug- 10child
‘Sipho will cook food for the children.’
b. u-Sipho
u- zo- thumela imali
ku- bantwana.
aug-1Sipho 1s- fut- send.appl aug.9money ku- 2child
‘Sipho will send money to the children’
Recall that there is no acc-marking on transitive nominals in Zulu and that the complement of the preposition in predicative possession does not behave like a transitive
object in many ways (triggering resumption under extraction, different object marking
patterns). At the same time, we see here that the object of the possessive preposition does
share underlying similarities with the transitive objects with respect to the distribution
of augments. The apparent transparency of na7 for the purposes of structurally-licensed
augment-drop is not shared by all other prepositions in the language, which instead seem
to simply replace the augment in all environments.
7 And
300
a few other prepositions in Zulu.
18 More on have and need
To summarize, the augment drop patterns in Zulu give us a test for possessees whose
results diverge from those of the tests in the previous section, grouping the complements
of na with transitive objects (and not with other prepositional complements). This discussion becomes particularly relevant for our concerns in light of Halpert (2012)’s proposal
that augment drop is only permitted in positions where structural Case is assigned, as
we discuss in the following subsection.
3.2 Case implications
The discussion in this section concerns the role of Case Theory in the Bantu language
family. While we do not present a definitive account of Case in these languages, we show
that the types of case-theoretic puzzles – and their proposed solutions – that emerge in
the Bantu family suggest that Harves & Kayne (2012)’s revised approach to have and
need in (10) may in fact be on the right track for these languages.
3.2.1 Existing proposals about Bantu case
Diercks (2012), building on a range of research (e.g. Harford Perez 1985; Ndayiragije 1999;
Alsina 2001; Baker 2003; 2008), showed that a wide variety of constructions crosslinguistically among Bantu languages – including raising constructions, locative inversion, and
possible-constructions, among others – do not behave in the familiar ways predicted by
Case Theory, two examples of which are included below: the first shows a perceptionverb raising construction that is equivalent of the ungrammatical English *John seems
that fell, in which the embedded subject has raised out of a tensed and agreeing clause,
where it presumably should have been Case-licensed and rendered inactive (known as
hyper-raising).
(23) Lubukusu hyper-raising
John a- lolekhana mbo ka- a- kwa.
1John 1s- seems
that 1s- pst- fell
‘John seems like he fell/John seems to have fallen.’
The example in (24), on the other hand, shows a noun phrase appearing as subject
of a non-finite clause where there is no evidence of a Case-licenser (overt or covert) to
license it.
(24)
Swahili overt subject of infinitive
I- na- wezakana (*kwa) Maiko ku- m- pig- ia Tegani simu.
9s- pres- possible (*for) 1Michael inf- o- beat- appl- fv 1Tegan 9phone
‘It is possible for Michael to call Tegan.’
Diercks concluded that these patterns indicate that Bantu languages simply lack abstract Case features, articulated in a macroparameter:
(25)
Case Parameter: Uninterpretable Case features are / are not present in a language.
301
Claire Halpert & Michael Diercks
Such a proposal raises the question of what (if any) prediction Harves & Kayne (2012)’s
revised generalization makes about have and need in languages without Case. One possibility, discussed below, is that absent Case, incorporation of need is unrestricted by the
absence of transitive HAVE.
Another approach to Case in Bantu emerged in the augment-drop discussion above.
As we saw, Halpert (2012) argues against parameterizing Case in Bantu, attributing augment distribution patterns to Case-licensing. Crucially, this Case-licensing system is distinct from standard nom-acc licensing patterns: Halpert argues that augments and some
prepositions give abstract Case to the nominals they mark and that abstract case is assigned to the highest element in vP. While it is unsurprising for prepositions to value
Case features, the claim about Zulu is that only certain prepositions do so (as illustrated
in the previous section). In addition, another surprising aspect of Halpert’s proposal is
that the augment, which is typically considered a DP-level prefix and not a preposition,
also licenses nominals. Nominals without valuation in these ways are restricted to structural Case positions, which again differ in a standard nom-acc language, where T and
v𝑜 are Case-licensers. Halpert proposes that Case is mediated by an intermediate phrase
(LP), which licenses downward to the highest element in vP, accounting for patterns
like those shown in §3.1. The result of this analysis is that Zulu Case, unlike acc, is not
connected to transitivity.
We do not attempt to resolve these differing approaches to Bantu Case here. Rather,
we point out that the consistent thread throughout all preceding work on this issue is
that Bantu Case is not business as usual. Whether one adopts a no-Case approach or a
non-nom/acc approach, we argue in the next section that both in fact predict a similar
pattern with respect to Harves & Kayne (2012)’s analysis of have and need.
3.3 Restating the generalization
We return now to the main problem that this Bantu data raises for Harves & Kayne
(2012)’s account: predicative possession shows non-transitive behavior, despite the existence of transitive need. As discussed in the previous section, multiple proposals suggest
that Case in Bantu is divorced from transitivity – either because there is no Case or because nominals are licensed by a projection above vP that is not linked to predicate type.
This consensus holds even if we don’t resolve the questions of Case-licensing in Swahili
or Zulu (or Bantu more broadly) here.
We propose that on either approach, the Case properties of transitive objects and
B-construction possessees are identical: either neither has Case, or both do (say, from
Halpert’s LP). Either way, this Case pattern is distinct from any traditional notion of accusative Case but uniform across predicative possession and transitive objects. The split
that we demonstrated in section 2 between behavior of possessees and direct objects in
syntactic tests for objecthood is expected because syntactic objecthood is divorced from
structural case on either account. What we have available to us, then, is a modification
of Harves & Kayne (2012)’s revised generalization in (10):
(26)
302
Need-Licensing Generalization:
All languages that have a transitive verb corresponding to need are languages in
which predicative possessees are licensed in the same manner as transitive objects.
18 More on have and need
Note that even if (26) accurately describes the current state of affairs, of course, it’s
not yet clear why it should be the case. The next subsection briefly discusses some ideas
in this vein.
3.4 Thoughts on the derivation of need
Recall that for Harves & Kayne (2012), the role of acc-assignment is closely tied to their
proposed derivation of need: Transitive need occurs when the theme of a nominal need
can get acc Case. If the nominal need incorporates to a transitive have, the acc from
have is available for its theme.8
(27)
VP
N+V
[need𝑖 + HAVE]
NP
t𝑖
DP
We’ve argued, however, that transitivity has nothing to do with how themes are licensed in Bantu. The question that arises, of course, is: if our revised generalization regarding the relationship between have and need holds, how does a Case-less derivation
of need fit into Harves & Kayne (2012)’s story? One possibility is that transitive need
can be built directly by incorporating the nominal need into the (non-transitive) copular
predicative possession construction. In other words, Harves & Kayne (2012)’s universal
derivation for transitive need breaks down for the Bantu languages discussed here precisely because (accusative) Case-licensing is de-linked from transitivity. Transitive have
is unnecessary for deriving need because the transitivity of have or HAVE is irrelevant
for the licensing of the object of need due to the different Case-licensing properties of
these languages. We suggest here, therefore, that transitive verbs can be derived from
non-transitive components in this type of language if the incorporating nominal has its
own theme: the theme can either be licensed by a higher head independent of transitivity of predicates (following Halpert 2012) or does not need to be licensed at all (Diercks
2012).
4 Conclusions
In this paper we have addressed recent discussion of the typological patterns surrounding the relationship between have and need. Following Antonov & Jacques (2014), we
have shown that the typological generalizations proposed by Harves & Kayne (2012) do
not hold up in the face of data from a variety of Bantu languages. Specifically, Zulu,
Swahili and Kuria all are B-languages with lexical need verbs, contrary to the proposed
generalization(s) of Harves & Kayne (2012). We moved beyond the evidence in Antonov
& Jacques (2014) to provide new syntactic tests that show that the Zulu and Swahili
counter-examples are in fact true counter-examples: B-possession is non-transitive while
8
Under the assumption that incorporated nouns don’t need Case, following Baker (1988)’s classic account:
if transitive have is unattested elsewhere in the language, there’s no base on which to build transitive need.
303
Claire Halpert & Michael Diercks
need patterns with other transitive verbs, ruling out the possibility that these languages
are somehow covert H-languages.
While Antonov & Jacques (2014) conclude on the basis of similar evidence that the
Harves & Kayne (2012) decompositional analysis of need is therefore incorrect and ought
to be abandoned, we investigated a potential alternative route. In light of recent research
suggesting deep differences between the properties of structural Case in Bantu and those
of the languages discussed in Harves & Kayne (2012)’s original survey, we proposed
the revised generalization in (26) that focuses not on have and need both assigning acc
case, but instead simply requires that have and need show the same structural licensing
properties.
This proposal gives us a new set of empirical predictions. Antonov & Jacques (2014) discussed several additional languages (Estonian, Moroccan and Algerian Arabic; Likpe and
Ewe from the Kwa family; and Ayacucho Quechua) that are surface counter-examples to
Harves & Kayne (2012)’s generalizations. If our proposal is on the right track, these languages ought to show similar licensing properties between predicative possession and
transitive need, even if they are not transparently related to acc case on the surface. We
see two potential outcomes of such investigations. The first is the same conclusion that
Antonov & Jacques (2014) arrive at: if there are not underlying similarities in the licensing of objects of B-languages with a transitive lexical need, then Harves & Kayne (2012)’s
generalization and our revised generalization proposed here may simply be inaccurate.
If so, both versions should be abandoned, suggesting that we may not want a universal
decomposition analysis of need after all.
Alternatively, we may find that the other exceptions to Harves & Kayne (2012)’s generalization noted by Antonov & Jacques (2014) are in fact rooted in underlying differences
in structural licensing, as we have proposed for the Bantu languages discussed here. If
this second possibility is borne out, then we may stand to uncover a deeper universal that
underlies Harves & Kayne (2012)’s initial observations. In particular, if the predictions
we discuss here are upheld, then Harves & Kayne (2012)’s generalizations (and our revision of them) point to some deep consistencies between languages (with with respect to
the decomposition of need) that can be obscured by differences between languages with
respect to structural licensing patterns. It is possible that this particular combination of
traits that is problematic for Harves & Kayne (2012) – B-languages with transitive need –
could ultimately be viewed as a diagnostic of underlying differences in structural licensing between Harves & Kayne (2012)’s languages and the ‘exceptional’ ones. These are
of course empirical questions, meriting additional empirical investigation, though with
potentially large theoretical import.
Acknowledgments
We would like to thank Johnes Kitololo, Justine Sikuku, Monwabisi Mhlophe, Mthuli Percival Buthelezi, and Tafadzwa Mtisi for lending their judgments to this project. Thanks
also to Stephanie Harves, Richard Kayne, and Rodrigo Ranero, the audience at ACAL 45,
and two anonymous reviewers for their helpful comments and suggestions.
304
18 More on have and need
References
Alsina, Alex. 2001. Is case another name for grammatical function? Evidence from object asymmetries. In William D. Davies & Stanley Dubinsky (eds.), Objects and other
subjects, 77–102. Dordrecht: Kluwer Academic Publishers.
Antonov, Anton & Guillaume Jacques. 2014. Transitive NEED does not imply transitive
HAVE: Response to Harves and Kayne 2012. Linguistic Inquiry 45(1). 147–158.
Baker, Mark C. 1988. Incorporation. Chicago, Illinois: University of Chicago Press.
Baker, Mark C. 2003. Agreement, dislocation, and partial configurationality. In Andrew
Carnie, Heidi Harley & MaryAnn Willie (eds.), Formal approaches to function in grammar, 107–134. Amsterdam: John Benjamins.
Baker, Mark C. 2008. On the nature of the Antiagreement Effect: Evidence from wh-insitu in Ibibio. Linguistic Inquiry 39(4). 615–632.
Bax, Anna & Michael Diercks. 2012. Information structure constraints on object marking
in manyika. Southern African Linguistics and Applied Language Studies 30. 185–202.
Buell, Leston & Merijn de Dreu. 2013. Subject raising in Zulu and the nature of PredP.
The Linguistic Review 30(3). 423–466.
Creissels, Denis. 2013. Control and the evolution of possessive and existential constructions. In Elly van Gelderen, Jóhanna Barðdal & Michela Cennamo (eds.), Argument
structure in flux. The Naples-Capri papers, 461–476. Amsterdam: John Benjamins Publishing Company.
Diercks, Michael. 2012. Parameterizing case: Evidence from Bantu. Syntax 15(3). 253–
286.
Halpert, Claire. 2012. Argument licensing & agreement in Zulu: MIT dissertation.
Harford Perez, Carolyn. 1985. Aspects of complementation in three Bantu languages. Madison, WI: University of Wisconsin-Madison dissertation.
Harves, Stephanie & Richard S. Kayne. 2012. Having ‘need’ and needing ‘have’. Linguistic
Inquiry 43(1). 120–132.
Kayne, Richard S. 2014. Bisogna, need, and is To. The syntactic character of lexical gaps.
Handout, 8th Cambridge Italian Dialect Syntax-Morphology Meeting, Università di
Padova. 15.
Marten, Lutz, Nancy Kula & Nhlanhla Thwala. 2007. Parameters of morpho-syntactic
variation in Bantu. Transactions of the Philological Society 105(3). 253–338.
Ndayiragije, Juvénal. 1999. Checking economy. Linguistic Inquiry 30(3). 399–444.
Nguako, Nduvaa Erna. 2013. The new otjiherero dictionary. Bloomington, Indiana: AuthorHouse.
Riedel, Kristina. 2009. The syntax of object marking in Sambaa: A comparative Bantu perspective. Leiden: Universiteit Leiden dissertation.
Zeller, Jochen. 2012. Object marking in Zulu. Southern African Linguistics and Applied
Language Studies 30. 219–325.
305
Chapter 19
Structural transfer in third language
acquisition: The case of Lingala-French
speakers acquiring English
Philothé Mwamba Kabasele
University of Illinois at Urbana-Champaign, University of Calgary, Institut Superieur
Pedagogique de la Gombe (RDC), University of KwaZulu-Natal
This paper tests the claims of Cumulative Enhancement Model, the ‘l2 status factor’, and the
Typological Primacy Model in investigating how l1 Lingala, l2 French speakers express in
English an event which took place and was completed in the past. The linguistic phenomena
understudy informs us that English uses the simple past in a past-completed event while
French and Lingala use the ‘passé composé’ and the remote or recent past, respectively. The
study circumscribes the tense similarities and differences between the three languages.
The paper strives to answer the questions on which previously acquired language between
the l1, l2, or both l1 & l2 overrides in l3 syntactic transfer. The paper aims to determine
whether the l2 is the privileged source of syntactic transfer even when the l1 offers syntactic similarities with the l3. Finally, the study purports to determine whether subjects are
more accurate when communicating in explicit mode than in implicit mode. That is, the
study further aims to investigate whether subjects make less transfer errors in a task that
promotes reliance on explicit knowledge than they do in task that promotes reliance on
implicit knowledge.
The findings of the study show that subjects used the simple past tense in the context of a
past-completed event. The use of the simple past tense in the context of a past-completed
event might be attributed to transfer from the l1 or might be considered as a consequence
of positive learning.
The results further show that subjects have transferred more explicit knowledge than implicit. And the results have ruled out the l2-status factor claim that the l2 is the privileged
source of transfer in l3 acquisition.
1 Rationale
This paper tests the claims of the Cumulative Enhancement Model (cem) by Flynn et al.
(2004), the ‘l2 status factor’ by Bardel & Falk (2007), and the Typological Primacy Model
Philothé Mwamba Kabasele. Structural transfer in third language acquisition: The case of
Lingala-French speakers acquiring English. In Jason Kandybowicz, Travis Major, Harold
Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected papers from
the 45th Annual Conference on African Linguistics, 307–323. Berlin: Language Science Press.
DOI:10.5281/zenodo.1251744
Philothé Mwamba Kabasele
(tpm) by Rothman (2010); Rothman (2011) in investigating how l1 Lingala l2 French
speakers express in English an event which took place and was completed in the past.
The work aims to test the claims of those three models of l3 acquisition in terms of
source of transfer and determine the factor, which takes precedence in determining the
source of transfer when there is the potential for competition between multiple factors.
The Cumulative Enhancement Model claims that previously acquired linguistic knowledge from both l1 and l2 positively impact the acquisition of any subsequent language.
The ‘l2 status factor’ privileges and restricts the source of transfer from only the l2 while
the Typological Primacy Model constrains transfer to the language that is perceived to
be (psycho)-typologically closer to the l3.
The paper studies the population of twenty-five Lingala speakers who also speak
French as l2 and who are learning English as l3, three languages of which two are IndoEuropean and one is a Bantu language. This is the first study, which combines those
three languages in the context of third language acquisition.
The linguistic phenomena under study inform us that English uses the simple past to
talk about an event which took place in the past and was completed in the past while
French and Lingala use the ‘passé composé’ and the past (remote or recent past), respectively. For the sake of this study, the simple past (historical past) in French is not
considered as a potential factor that can trigger transfer because as Rowlett (2007: 26)
argues that changes in the spoken language in French have taken place in the use of the
passé composé’ in which the perfect is used instead of the past historic when talking
about past completed event. Furthermore, with reference to the economy of cognitive
design and linguistic architecture (Flynn et al. 2004; Rothman 2010) and in relation to
the biological theory of language acquisition (Chomsky 2007), it is observed and documented that speakers of a language prefer the most economical linguistic option and this
preference is hardwired into human cognition or better in the grammar of the language
(Rothman 2010: 271). It is postulated in this paper that the l2-speaking French subjects
would resort to the ‘passé composé’ to talk about a past completed event rather than the
historical past (simple past) because the former is the option that is available to them and
the parser would strongly and straightforwardly prefer the option which offers easier
access.
Both French and English present some syntactic similarities in terms of form while
they differ in terms of function. Their similarity is observed between the form of the
‘present perfect tense’ and the form of the ‘passé compose’ which are structured as ‘aux
have/avoir + past participle’ in both languages. The differences are observed in their
function; the ‘passé composé’ is used in French to talk about an event which took place
and was completed in the past while the ‘present perfect tense in English expresses an
event that started in the past but has some implication in the present. Whereas, Lingala
and English show some syntactic similarities in terms of both form and function because
the simple past in both languages are used to talk about a past completed event and in
terms of form both languages use inflectional morphemes to morpho-syntactically mark
the past tense. The study circumscribes the similarities and differences, in tense, between
the three languages.
308
19 Structural transfer in third language acquisition
The paper strives to answer the questions on which previously acquired language
between the l1, l2, or both l1 & l2 overrides in l3 syntactic transfer. The paper aims to
determine whether the l2 is the privileged source of syntactic transfer even when the l1
offers syntactic similarities with the l3. Finally, the study purports to determine whether
subjects are more accurate when communicating in explicit mode than in implicit mode.
That is, the study further aims to investigate whether subjects make less transfer errors in
a task that promotes reliance on explicit knowledge than they do in task that promotes
reliance on implicit knowledge. The predictions of the study permit to test the three
models, specifically to test the descriptive and explanatory adequacies of the cem, ‘l2
status factor’, and tpm.
Apart from the rationale and the conclusion, §2 is on the review of literature. §3 discusses the linguistic phenomenon that motivates the study. §4 provides the predictions
and research questions of the study. §5 presents the design and methodology of the study.
§6 is on the results while §7 presents the discussion on the findings of the paper.
2 Literature review
2.1 Transfer in L3 acquisition
Transfer in l3 varies depending on the domain and two languages are identified as source
of transfer; the l1 or the l2. When a learner assumes that his l1 is closely related to the
tl, there is a high likelihood for the l1 to trigger more transfer than when he assumes the
opposite. Clearly, it is more probable to have less transfer when a learner assumes great
linguistic distance between, say, his l1 and the tl in l3 acquisition. The speaker’s perception of language similarity, which is psychological and does not necessarily reflect the
actual linguistic distance between the languages, may trigger or constrain transfer in the
acquisition of l3. Ringbom (2003) has restricted the importance of perceived typological
distance in the transfer of lexis. When l2 and l3 offer a considerable number of common
cognates, the speaker perceived both languages as similar and this psychotypological
effect favors transfer.
2.2 Syntactic models in L3 acquisition
The three syntactic models in l3 acquisition agree upon the influence of, at least, one
previously acquired language. They, however, depart from one another by the way they
formulate their predictions.
2.2.1 Cumulative enhancement model (CEM)
The cem (Flynn et al. 2004) claims that language learners rely on both their l1 and l2
cumulated linguistic knowledge when acquiring an additional language. This claim identifies language acquisition in a multilingual context as a cumulative process. The multilingual learner’s reliance on the previously acquired linguistic knowledge is restricted
to only transfer which has a noticeable rewarding impact in the learning process of the
309
Philothé Mwamba Kabasele
subsequent language. The previously acquired languages can positively contribute to the
acquisition of a l3. The insistence of cem on the sole beneficial effects of previous linguistic knowledge in the acquisition of an additional language implies a denial of negative
transfer from previously acquired languages.
Flynn et al. (2004) ascertained that, “Language acquisition has a scaffolding effect”
(Rothman 2010: 110). This means any previously acquired linguistic knowledge’s role
is twofold. It can either enhance the acquisition of any additional language or remain
neutral. The impact of both l1 and l2 in the process of the acquisition of an additional
language is relevant. The l2 contribution only supersedes that of l1 when, say, structure
wise, the syntactic features which are in play are not available in the l1 linguistic system.
2.2.2 The L2 status factor
The ‘l2 status factor’ (Bardel & Falk 2007) privileges the l2; it argues that the l2 is the
only linguistic system, which imposes its features onto the subsequent language. Bardel
& Falk (2007) claim that the acquisition of the l3 is qualitatively different from those
of the previously acquired languages because the linguistic knowledge of l2 plays a
substantial role in facilitating the process (see also Hufeisen 1998; Cenoz & Jessner 2000;
Cenoz 2001; 2003).
The claim, that the l2 is the strongest source of transfer in l3 acquisition stems from
the findings of the studies by Bardel & Falk (2007) and Falk & Bardel (2011). The findings
of this study were congruent with the claim that l2 is the strongest source of initial
transfer in l3. In their recent paper, Falk & Bardel (2011), they studied the placement of
object pronouns and their findings confirmed the privileged role of l2 in acquiring an
l3.
2.2.3 Typological primacy model (TPM)
Rothman (2011: 233) stipulates, “Initial State transfer for multilingualism occurs selectively, depending on the comparative perceived typology of the language pairings involved or psychotypological proximity.” The model argues that typological proximity or
psychotypology constrains transfer to the l3. The prevailing role of typological similarity and its role as a crucial variable in the acquisition of an l3 are significant.
Transfer does not always happen in a facilitative fashion. tpm predicts that in a pair
of previously acquired languages only the one, which offers typological proximity with
the target language, serves as the source of transfer. tpm constrains transfer from two
perspectives: the actual typological proximity or the perceived typological proximity,
which is also called psychotypological proximity existing between the three grammars
(García-Mayo & Rothman 2012: 19).
García-Mayo & Rothman (2012: 19) argue that, “At the initial state upon a limited
amount of exposure to the target l3, the tpm proposes that the internal parser assesses
relative typological proximity and selects which system should be transferred.” The tpm
is selective and conditionally non-facilitative. The parser selects the closest system to
the tl. Any syntactic feature such as word order, tense similarity, or any other syntactic
310
19 Structural transfer in third language acquisition
similarity depending on the case that is observed at the syntactic level may determine
the selectivity of one of the competing previously acquired languages.
The criticism that is formulated against tpm addresses its apparent incapacity to predict the source of transfer when the languages at hand do not present any clear typological proximity. The tpm suggests that transfer can be non-facilitative when psychotypology conditions the transfer by matching and misanalysing the underlying syntax of l1
or l2. Should it be noted that tpm is not clear on the interpretation of ‘typology’? The
term is unclear and it lends itself to ambiguous interpretation. One can interpret it as
referring to a specific linguistic structure, which is otherwise called for the sake of this
paper ‘local typology’; or one can also interpret it as referring to the whole language,
which is called ‘global typology’. For the sake of clarity in this study, I refer to typology
as a specific linguistic structure or local typology.
3 The linguistic phenomenon
This section discusses and contrasts the use of the past tense, simple present, and the
present perfect tense in the three aforementioned languages. The simple past exists in
French, English, and Lingala while the form, “Aux (have/avoir) + past participle” exists
only; form wise, in both French and English.
The present perfect tense in English is made up of the auxiliary “have” plus the past
participle. This tense is similar in form to ‘passé composé’ in French, which is also made
up of the auxiliary “avoir” (have) plus the past participle. The present perfect and the
‘passé composé’ tenses present the same formal paradigm but differ in terms of their
function.
The present perfect tense is always used in English to talk about a past until now
event while the ‘passé composé’ in French is often used to express a past-completed
event. Syntactic change has taken place in French in which the ‘passé composé’ is used
instead of the past historic to express a past-completed event. At this point, I can claim
that the English present perfect tense is similar to French ‘passé composé’ with respect
to form but it does not exist in Lingala. Therefore, different tenses are used to express
the same idea but in a different language. For instance, in English, the present perfect
tense is used with expressions like ‘the first time’, ‘the second time’, since, for, etc., while
in French and Lingala the present and the immediate past are respectively used.
The simple past tense is used in English to talk about events which took place in the
past and when the time period is completed. In French and Lingala the ‘passé composé’
and the past are respectively used. In Lingala, an appropriate past tense form needs to
be selected depending on whether the event was completed in the recent past or in the
distant past. Example (1) illustrates the case.
(1)
English simple past
Joe bought a car last year.
311
Philothé Mwamba Kabasele
(2)
French passé composé
Joe a acheté une voiture l’ année passé.
Joe has bought a car
the year past
‘Joe bought a car last year.’
(3)
Lingala recent past
Joe asomb -aki mutuka mbula eleki.
Joe 1pssva buy rec.pst car
year past
‘Joe bought a car last year.’
Tables 1 and 2 summarize tenses in these three languages.
Table 1: Past event that was completed in the past.
Tense
Example
English
Simple past
Andy went to Paris last
month.
French
Passé composé
Andy est parti à Paris le
mois passé.
‘Andy went to Paris last
month.’
Lingala
Remote past
Andy akendáká na Paris
bambula eleka.
Andy akendaki na Paris
sanza eleki.
‘Andy went to Paris years
ago.’
‘Andy went to Paris last
month.’
Recent past
Gloss
Table 2: Past event with connection in the present.
Tense
Example
Gloss
English
Present perfect
Nathan has lived in
Urbana since 2011.
French
Indicatif present
(Simple present)
Nathan vit à Urbana
depuis 2011.
‘Nathan has lived in
Urbana since 2011.’
Lingala
Past (Immediate past)
Nathan afandi na
Urbana banda 2011.
‘Nathan has lived in
Urbana since 2011.’
In this paper, my attention is first focused on past-completed event whereby the simple past is used in English while the ‘Passé composé’ and remote/recent past are used
312
19 Structural transfer in third language acquisition
in French and Lingala respectively. The linguistic phenomenon which is the focus of the
predictions in this study informs us that both the ‘present perfect’ in English and the
‘passé composé’ in French present form similarities in terms of their syntactic structure
which is “the auxiliary have + the past participle” but they diverge in terms of their function. The ‘passé composé’ is used to express a past-completed event while the ‘present
perfect tense’ is used to express a past until now event. The ‘passé simple’ in French will
not transfer because it is marked and is hardly used in spoken communication; it is a
tense that is used by highly educated people in literary discourse.
Second, my interest is oriented to past until now event in which the present perfect
tense is used in English while for the same event, French uses the simple present tense
but Lingala uses the (immediate) past.
4 Predictions and research questions
The predictions of this paper are organized as such: Based on the tpm which claims that
only the language with syntactic proximity with the tl serves as the source of transfer,
the study posits that if subjects are tapping their linguistic knowledge from the l1 to
talk about a past completed event in English they will use the simple past tense. This
tense choice will be triggered by the local syntactic similarity in terms of form between
the simple past in English and the remote/ recent past in Lingala in the context of a
past-completed event.
Transfer occurs because subjects make an interlingual identification; they perceive
and judge that the form of the syntactic structure of the remote/recent past in Lingala
is similar to the form of the syntactic structure of the simple past tense in English. It is
also the form of the syntactic structure of the simple past in English, which has invited
the perception of the similarity between the forms of the sentences in both languages.
Transfer is triggered by the psychotypological constraint, which enables subjects to perceive similarity between the two tenses. This similarity is observed at the level of form
of the tenses. It is hence clear that the syntactic structure of a previously acquired language is susceptible to transfer as Jarvis (2010: 174) puts it, “only if it is perceived to
have a similar counterpart in the recipient language.” The perception of the similarity is
not only observed on the surface level but subjects’ perception of the similarity at the
psychological level plays also a role for transfer to occur.
With reference to the “l2-status factor” model which claims that the l2 is the strongest
source of transfer in l3 acquisition and that the l2 blocks any syntactic transfer from the
l1 syntactic system, the study posits that if subjects are tapping their linguistic knowledge from the l2 to talk about a past completed event in English they will use the present
perfect tense. As discussed earlier, the ‘passé simple’ in French will not transfer because
it is marked and is hardly used in spoken communication. It is a tense that is used by
highly educated people in literary discourse and it has been replaced by the ‘passé composé’.
Based on the cem which claims that learners rely on their cumulated linguistic knowledge from both l1 and l2 as source of transfer and that transfer is only positive or null;
313
Philothé Mwamba Kabasele
the study posits that if subjects are tapping their linguistic knowledge from both l1 and
l2 to talk about a past completed event in English they will use the simple past tense.
In light of the decisive factors, closeness between the l1 Lingala and l3 English (in
form) but difference in ‘form’ between the l2 (passé compose) and l3 (simple past)
and the aforementioned predictions, the work seeks to answer the following questions:
Which previously acquired language between the l1, l2, or both l1 & l2 takes precedence in l3 syntactic transfer? Is the l2 the privileged source of syntactic transfer even
when the l1 offers some syntactic similarities with the l3? Answers to these concerns
will shed light on my study.
5 Design and methodology
5.1 Subjects
Twenty-five adult Congolese immigrants who live in the USA participated in the study.
The average age when they started to be exposed to English is 15 years old and their
average length of residence in the USA is 3 years. Most of them acquired French through
instructional exposure at school and their average length of exposure through formal
instruction in French is 4 years. They also formally learned English as a school subject.
All the subjects grew up in Kinshasa and attended school in the same setting. They are
all native speakers of Lingala who also speak French as an official language. The latter is
used as an official language and as the language of instruction from elementary school
upward. French was also learned as a school subject whereby emphasis was made on
the grammar of French. English was exclusively learned as a school subject. Subjects
started taking English from ninth grade of high school up to twelfth grade. However,
English was heavily taught structurally. Little attention was paid to other language basic
skills. Therefore, students completed the high school program with very poor speaking,
reading, writing, and listening skills. They all have at least a high school state diploma
from the Democratic Republic of the Congo. Subjects are, however, exposed to English
in the USA on a daily basis at work place, stores, and public places. They tend to interact
in Lingala whenever they meet with other Congolese fellows.
Subjects with advanced level of proficiency in English could hold a long conversation
of approximately ten minutes in English. They exhibited oral fluency but with a few
grammatical errors. They could ask clarification questions and could answer questions
on social life topics with certain ease. Intermediate proficiency level subjects could ask
questions with hesitation and were able to hold an intelligible interaction despite some
observed limitation they displayed in vocabulary. They made random language errors,
which sometimes could lead to communication breakdown. Beginner subjects had limited English proficiency level. They had significant amount of difficulty in the four basic
language skills. They were hesitant in their speech and their grammar was poor. Subjects
were administered a cloze test to determine their proficiency level in English.
The control group was made up of five American native speakers of English. All the
five subjects grew up in the USA and have been exposed to English since birth. They all
have at least a high school degree and had taken at least a foreign language at school.
314
19 Structural transfer in third language acquisition
5.2 Task and procedure
I administered the interview and the written elicitation task to the subjects to collect the
data of the study. The interview was always administered prior to the written elicitation
task an each subject took both tasks on the same day. Subjects were interviewed in English and the interview, which was audio recorded elicited the data through oral mode of
interaction and under time pressure. I used the smartphone Alcatel one touch to record
the interviews. Whereas, the written elicitation task aimed to elicit data through the written mode of communication and allowed enough time to subjects to express themselves.
Obviously, the interview elicited the data through the implicit mode while the written
elicitation task did so through the explicit mode. All the questions were used only once
in each task. They were never repeated in another task. All the interviews were recorded
and then transcribed.
The interview was related to past-completed event. It aimed to elicit verb tense forms
in the simple past (questions 1 and 3). The future (question 4) was used as a distractor in
the study. The questions aimed to trigger a specific verb tense in the speech production
of each subject. The simple past category had two questions while the future category
had only one question. The question related to the future was a distractor. For the sake
of this study, after analysis of the questions, only questions 1 and 3 were reported. Data
related to question 2 would be incorporated in the larger project, which is related to
this study. Question 4 was not reported because it was a distractor. The following are
the interview questions: Tell me about something that you remember from your life in
Congo? Tell me about your two big accomplishments in the last six months? Tell me
about your first arrival in the USA? Tell me about something that you would like to do
in six months? All the four interview questions were asked in the same order prior to
handing out the written elicitation task to the subjects.
Later on in the analysis of the interview, three coders determined the obligatory contexts in which the simple past tense had to be used. I was the first coder. Then, two other
coders who were native speakers of English contributed with their expertise. The native
speakers were teachers of English who were trained as teachers of English to speakers
of other languages or linguists. The coding was first done separately. And then, all the
three coders came together to discuss some minor differences, which were observed.
However, the written elicitation task had 24 questions. The task was organized into a
category of six items. The targeted category was the simple past tense and the present
perfect tense; the future, the simple present and the present progressive were distractors.
In this study, only the category of items that are related to the use of the simple past tense
are reported and data related to the use of the present perfect tense will be reported in
the other parts of the whole projects. All the instructions were given in English.
5.3 Proficiency test
The proficiency of the subjects was determined through the administration of the cloze
test. The cloze test was an adaptation from the American Kernel Lessons that was drawn
from the Advanced Students’ Book by O’Neill et al. (1981). The cloze test provided blanks
with three options of which subjects had to choose one in order to fill in the blank space
315
Philothé Mwamba Kabasele
with the option s/he deemed as the correct answer. The results of the test divided the
subjects into three proficiency groups: beginning, intermediate, and advanced levels with
the scores varying between 18 to 24, 25 to 29, and 30 to 37 respectively. Beside the cloze
test, subjects were also administered the language learning history in order to elicit their
language learning background, their personal data, and the family linguistic history.
6 Results
Discussing the results of the paper, the first task was the interview while the second
was the written elicitation and the results are presented in tables, which quantify the
former with respect to the category of items. The columns present the required context
in which a given tense was expected to be used (this is the target tense), the prediction(s)
to the category of items, i.e., the various tenses which were predicted, and finally the
unexpected answers which were called “Other verbal forms”.
The inferential statistics was conducted to compare the control group’s use of the
simple past and present perfect tense with the 3 proficiency groups that is, beginner,
intermediate, and advanced groups in the context of past-completed event. Its goal was
to determine whether the control group’s use of the simple past and present perfect
tense in the aforementioned contexts was significantly different from that of beginner,
intermediate, and advanced groups respectively. Moreover, it also aimed to help draw
sound decisions therefore on whether the use of the simple past and the present perfect
tense by the 3 proficiency groups could be attributed to transfer or not.
Although the goal of this study was to examine the kinds of forms that subjects used in
different circumscribed contexts rather than merely focusing on comparing the different
groups in the study, I hope that inferential statistics will also contribute in inducing
sound decisions on the interpretation of the results.
Table 3: Response types to interview eliciting the context of past completed
event (Task 1).
Simple past
Beginner
Intermediate
Advanced
Control
Present
perfect
Simple
present
Other
verbal forms
N
%
N
%
N
%
N
%
66
140
162
41
64.7
79.0
92.5
95.3
0
0
0
2
0
0
0
4.6
36
35
13
0
35.2
19.7
7.4
0
0
2
0
0
0
1.1
0
0
A one-way anova was conducted to compare the control group with the 3 proficiency
groups with respect to the use of the simple past tense in the context of past-completed
event whereby the use of the simple past tense expressed in percentage was the dependent variable and the groups the independent variables. The anova reveals that there
316
19 Structural transfer in third language acquisition
100
80
1.1
7.4
4.6
92.5
95.3
19.7
35.2
60
40
79
64.7
20
0
Beginner
Simple Past
Intermediate
Present Perfect
Advanved Control Group
Simple Present
Other verbal forms
Figure 1: Interview eliciting the context of past completed event.
were no significant differences between the control group and the 3 proficiency groups
[F (3, 29) = 2.36, p=.094]. A word of caution should be mentioned that given the small
sample size in this study, I suspect that the small sample size might have impacted the
statistical power to reach the significant difference between the control group and the 3
proficiency groups.
Table 4: Response types to the written elicitation task on past completed event
(Task 2).
Simple past
Beginner
Intermediate
Advanced
Control
Present
perfect
Simple
present
Other
verbal forms
N
%
N
%
N
%
N
%
40
49
41
30
74
90.7
97.6
100
0
4
1
0
0
7.4
2.3
0
2
0
0
0
3.7
0
0
0
12
1
0
0
22.2
1.8
0
0
A one-way anova was conducted to compare the control group with the 3 proficiency
groups with respect to the use of the simple past tense in the context of past-completed
event whereby the use of the simple past tense expressed in percentage was the dependent variable and the groups the independent variables. The anova reveals that there
317
Philothé Mwamba Kabasele
100
1.8
7.4
22.2
80
2.3
3.7
60
40
97.6
90.7
74
100
20
0
Beginner
Simple Past
Intermediate
Present Perfect
Advanved Control Group
Simple Present
Other verbal forms
Figure 2: Written elicitation task on past completed event.
were no significant differences between the control group and the 3 proficiency groups
[F (3, 29) =2.17, p=.11]. As mentioned earlier, the small sample size in this study might
have impacted the statistical power to reach the significant difference between the control group and the 3 proficiency groups.
7 Discussion
Both the tpm and cem predicted that subjects will use, to talk about past completed event,
the simple past tense as a result of respectively transfer from the l1 which shows syntactic proximity with the tl and transfer from both the l1 and l2 as a result of cumulated
knowledge. The “l2- status factor” predicted that subjects will use the present perfect
tense in the aforementioned context.
In the present study, I investigated structural transfer in third language acquisition. In
the interview and written elicitation tasks dealing with past-completed event the results
showed that subjects used the simple past tense in the context of past-completed event.
The results are contrasted with the predictions of the study in Table 5.
Referring to the context of past-completed event, the results raise the question of
knowing whether the use of the simple past tense by the subjects in the context was
due to transfer from the previously acquired languages or whether it was the result
of successful acquisition of the tense in the l3. The results of the inferential statistics,
which I take with reserve, in relation to the use of the simple past tense in the context
of past completed event in both tasks, that is, the interview and the written elicitation
318
19 Structural transfer in third language acquisition
Table 5: Predictions and results of the study.
Context
l1 Transfer
l2 Transfer
l1 & l2 Transfer
English
Past completed
event
Simple past
(tpm)
Present perfect
(l2 status factor)
Simple past
(cem)
Simple past
task fairly shows that there were no significant differences between the control and the
3 proficiency groups. In both the interview and the written elicitation task, the anova
showed respectively that there was no significant differences between the control group
and the 3 proficiency groups [F (3, 29) = 2.36, p= .094] and [F (3, 29) = 2.17, p= .11].
However, because of the small sample size of the study, the inferential statistics is not
taken into consideration because I suspect that the small sample size of the study might
have affected the statistical power to reach the significant difference between the control
group and the 3 proficiency groups, yet numerically the difference between those groups
are obvious.
Referring to the descriptive statistics, specifically to the numerical results as they are
depicted on table 5 and 6, there seem to be obvious differences between the control
group and the 3 proficiency groups. It is likely that subjects are tapping their linguistic
knowledge from the l1 to express in the tl an event, which took place in the past and
was completed in the past. However, the possibility of interpreting the results as a consequence of learning from the input is still open because if the use of the simple past
tense was solely attributed to transfer effects, we could expect to have more transfer
with beginners than with advanced proficiency groups.
Considering the transfer option, the results suggest that when an l1 offers some syntactic similarities with the tl, its (l1) syntactic system becomes transparent and thus accessible to the learners. This finding challenges the claims of the l2 status factor, which
postulate that the l2 blocks, the access to the syntactic system of the l1. I assume that
the l2 blocks access to the l1 syntactic system only when the latter does not display any
similarities with the syntactic system of the tl.
Should it be mentioned that it is not clear whether transfer from the l1 was due to
the effects of previously cumulated linguistic knowledge or just a matter of syntactic
proximity which was observed between the two linguistic systems. With reference to
the numerical results on the aforementioned tables, I suspect that l1 transfer into the
tl in this study was triggered by the syntactic proximity. The great number of simple
past tense use by advanced learners in the context of past-completed event shows that
there was positive transfer or positive learning as I will discuss it later in this section.
However, the high use of the simple present tense by beginners at the rate of 35.2% and
by intermediate learners at the rate of 19.7% implies that those learners are using the
simple past tense but they just fail to inflect the verb with the appropriate past tense
inflectional morpheme. The proficiency factor boosts and ameliorates the access to the
syntactic system of the l1.
319
Philothé Mwamba Kabasele
The other reading of the results attributes subjects’ performance to learning. It is likely
that the use of the simple past tense in the past-completed context may be due to learning.
It might further be interpreted that subjects successfully learned the use of the simple
past tense in past completed event context and that the occurrence of simple present
tense use in this context might be just attributed to failure to append the simple past tense
inflectional morpheme to the verb stem since subjects have not mastered the morphology
inherent to the simple past tense yet.
Furthermore, contrasting their performance in interview versus written elicitation
task with reference to subjects’ use of the simple present tense in the context of past
completed event, it is observed that in the interview whereby subjects had to resort to
their implicit knowledge due to time pressure they made more omission errors than
in the written elicitation task which required explicit knowledge. The rate of omission
errors was decreasing and correlated with the level of proficiency: beginners 35.2%, intermediate 19.7%, and advanced 7.4%. Whereas, in the written elicitation task, beginners’
rate of omission errors was relatively low, i.e., 3.7% while intermediate and advanced
subjects did not make any omission error at all. The type of knowledge one resorted to
can account for this difference. In the interview, subjects did not have enough time to
think and readjust their speech as they were being interviewed while in the case of written elicitation task, subjects had more time to prepare their answers and to observe that
there was an inflectional morpheme missing and they could self-correct their mistakes
by appending the omitted simple past tense inflectional morpheme to the verb form.
Figure 1 depicts, in a stairs-like manner, how the use of the simple past tense correlates
with the level of proficiency. Inversely, it also depicts how the occurrence of the simple
past tense inflectional morpheme omission errors correlates with the same level of proficiency. This reinforces the option that subjects are at a learning stage whereby they have
learned that the simple past tense should be used in the context of past-completed event
but they are still struggling with inflecting the verb with the appropriate morphological
marker, which will express and mark the simple past tense.
The use of the simple present tense in this context could be justified as the result of
error of inflectional morpheme omission. This could imply that subjects made positive
transfer but just failed to appropriately inflect the verbs in the past tense. Subjects need
more time to reinforce the learning of function/use of the simple past tense which seems
to be acquired but mostly to digest and control the appropriate morphological form to
append in order to fully acquire the tense.
With reference to the prediction related to past-completed event, the use of the simple past tense by the subjects is the result of positive transfer. The use of the simple
present tense in this context is considered as the result of error of the simple past tense
inflectional morpheme omission.
In light of the research questions which sought to determine the language that takes
precedence as source of syntactic transfer in l3 acquisition, the research question which
aimed to determine whether the l2 syntactic system blocks the syntactic transfer even
when the l1 offers some syntactic similarities with the l3, the interpretation of the results could be twofold. With reference to the inferential statistics, the latter did not have
enough statistical power to determine the difference between the control group and
320
19 Structural transfer in third language acquisition
the three proficiency groups. The statistical power was affected and weakened by the
small sample size of the subjects. However, because of the small sample size in the study,
which might have affected the statistical power, one can consider looking at the descriptive statistics, particularly the numerical results as they are presented on table 5 and 6.
Numerically, it is obvious that there was transfer. Responding to the question ‘Which
previously acquired language between the l1, l2, or both l1 & l2 takes precedence in l3
syntactic transfer’ the answer would be that transfer came from the l1.
Answering the second question which aimed to determine whether the l2 was the
privileged source of syntactic transfer even when the l1 offers some syntactic similarities
with the l3, the answer would be no. The l2 does not serve as the privileged source of
transfer when the l1 offers syntactic similarity with the l3.
Finally, attempting to answer the question of knowing whether subjects have more
and easy access to their implicit knowledge than the explicit knowledge and therefore
transfer more explicit knowledge than the implicit one when tapping linguistic knowledge from a previously acquired linguistic system, the results have shown that subjects
are more accurate when given the opportunity to use their explicit knowledge. This finding corroborates with those of previous studies whereby it was attested that subjects
were more accurate when in explicit mode than in implicit one (Schmidt 2001; 1995;
Leow 1998; Robinson 1997). It should, however, be noted that the erroneous use of the
simple present tense in the context of past completed event was mostly observed in the
context of implicit task. This shows and might imply that learners are linguistically unsecured when in implicit mode and thus they become inaccurate when they rely upon
implicit knowledge in their use of the target language.
8 Conclusion
The findings of the study attribute the use of the simple past tense in the context of pastcompleted event to positive transfer. However, the possibility of attributing the results
to positive learning is also to consider since inferential statistics did not reach any significance differences. The findings of inferential statistics were discarded because they
were affected by the small sample size and thus could not determine significant difference between the control and the 3 proficiency groups.
The use of the simple present tense, in the context of past-completed event, which
cannot be accounted for by transfer in this context is the result of omission of the simple
past tense inflectional morpheme. This failure by the subjects to append the simple past
inflectional morpheme to the verb to express the simple past tense shows that subjects
have not fully acquired the morphology inherent to the simple past tense and this was
mostly observed in implicit task.
The study has further shown that subjects were more accurate in using their explicit
than implicit knowledge. They also made more positive transfer from explicit knowledge
than from implicit one. The use of the simple past tense in the past completed event
context could possibly suggest that the learners only know one way to discuss events
that happened in the past using the simple past.
321
Philothé Mwamba Kabasele
I envisage replicating this study with a representative number of subjects in order to
avoid any negative implication on the statistical power. I will integrate the comprehension aspect of language transfer to have a full understanding of both production and
comprehension. I further project to present a hierarchical matrix of potential factors
which can trigger transfer and rank them in pairs, triplet or in quadruplet depending on
the factors which will be controlled.
References
Bardel, Camilla & Yehuda Falk. 2007. The role of the second language in third language
acquisition: The case of Germanic syntax. Second Language Research 23. 459–484.
Cenoz, Jasone. 2001. The effect of linguistic distance, l2 status and age on cross-linguistic
influence in third language acquisition. In Jasone Cenoz, Britta Hufeisen & Ulrike Jessner (eds.), Cross-linguistic influence in third language acquisition: Psycholinguistic perspectives, 8–20. Clevedon: Multilingual Matters.
Cenoz, Jasone. 2003. The additive effect of bilingualism on third language acquisition: A
review. International Journal of Bilingualism 7. 71–87.
Cenoz, Jasone & Ulrike Jessner (eds.). 2000. English in Europe: The acquisition of a third
language. Clevedon: Multilingual Matters.
Chomsky, Noam. 2007. Of minds and language. Biolinguistics 1. 9–27.
Falk, Yehuda & Camilla Bardel. 2011. Object pronouns in German L3 syntax. Evidence
for the L2 status factor. Second Language Research 27. 59–82.
Flynn, Suzanne, Claire Foley & Inna Vinnitskaya. 2004. The cumulative-enhancement
model for language acquisition: Comparing adults’ and children’s patterns of development in first, second and third language acquisition. International Journal of Multilingualism 1. 3–17.
García-Mayo, Maria P. & Jason Rothman. 2012. L3 morphosyntax in the generative tradition: The initial stages and beyond. In Jennifer C. Amaro, Suzanne Flynn & Jason
Rothman (eds.), Third language acquisition in adulthood, 9–32. Amsterdam: John Benjamins.
Hufeisen, Britta. 1998. L3 – Stand der Forschung – Was bleibt zu tun? In Britta Hufeisen &
Beate Lindemann (eds.), Tertiärsprachen. Theorien, Modelle, Methoden, 169–183. Tübingen: Stauffenburg.
Jarvis, Scott. 2010. Comparison-based and detection-based approaches to transfer research. In Leah Roberts, Martin Howard, Muiris Ó Laoire & David Singleton (eds.),
EUROSLA Yearbook 10, 169–192. Amsterdam: Benjamins.
Leow, Ronald P. 1998. Toward operationalizing the process of attention in second language acquisition: Evidence for Tomlin and Villa’s (1994) fine-grained analysis of attention. Applied Psycholinguistics 19. 133–159.
O’Neill, Robert, Edwin T. Cornelius & Gay N. Washburn. 1981. American kernel lessons,
AKl: Advanced. New York: Longman.
322
19 Structural transfer in third language acquisition
Ringbom, Hakan. 2003. If you know Finnish as L2, there will be no major problem learning
Swahili. Paper presented at the International Conference on Trilingualism and Third
Language Acquisition, Tralee, Ireland, September 4-6.
Robinson, Peter. 1997. Generalizability and automaticity of second language learning
under implicit, incidental, enhanced, and instructed conditions. Studies in Second Language Acquisition 19(2). 223–247.
Rothman, Jason. 2010. On the typological economy of syntactic transfer: Word order and
relative clause attachment preference in L3 Brazilian Portuguese. International Review
of Applied Linguistics 48. 245–274.
Rothman, Jason. 2011. L3 syntactic transfer selectivity and typological determinacy: The
typological primacy model. Second Language Research 27. 107–127.
Rowlett, Paul. 2007. The syntax of French. Cambridge: Cambridge.
Schmidt, Richard. 1995. Consciousness and foreign language learning: A tutorial on the
role of attention and awareness in learning. In Richard Schmidt (ed.), Attention and
awareness in foreign language learning (Technical Report 9), 1–64. Honolulu: University of Hawaii.
Schmidt, Richard. 2001. Attention. In Peter Robinson (ed.), Cognition and second language
instruction, 3–32. Cambridge: Cambridge University Press.
323
Chapter 20
Adjectives in Lubukusu
Aggrey Wasike
University of Toronto
The lexical category of adjectives is proposed to be universal, but its realization varies across
languages. In languages such as English, there is a clearly distinct category of adjectives. But
in other languages the category of adjectives is not entirely distinct morphologically and
syntactically from nouns and verbs. In this paper I show that there is a striking resemblance
between adjectives and nouns in Lubukusu. In addition, stage-level predicate meanings are
expressed by use of verbs rather than adjectives. Because of these facts, it is tempting to
adopt an analysis that reduces Lubukusu adjectives to either nouns or verbs. However, I
argue that there is not sufficient evidence to support such an analysis. Lubukusu has true
adjectives in spite of the associated nominal and verbal characteristics. A verbal characteristic such as expressing adjectival meanings by use verbs is similar to languages such Mohawk
and Vaeakau-Taumako. But there are significant differences between these languages and
Lubukusu with regards to this verbal characteristic.
1 Introduction
This paper discusses adjectives in Lubukusu, a Bantu language spoken in Western Kenya.
This discussion is particularly useful considering the fact that there is disagreement
among linguists on the status of adjectives as a universal category. For linguists such
as Mark Baker and R. M. W. Dixon, the lexical category of adjectives is universal (Baker
2001; 2003; Dixon & Aikhenvald 2004). But for R. M. W. Dixon in his earlier work and
Chafe Wallace, the category of adjectives is not universal (Dixon 1982; Chafe 2012). Description and discussion of how adjectival meanings are expressed in different languages
can help linguists draw a valid conclusions regarding the category of adjectives.
The lack of agreement on the status of adjectives as a universal category among linguists is attributed to cross-linguistic variation in the expression of adjectival meanings.
Some languages express adjectival meanings by use of adjectives while others express
similar meanings by use of verbs and nouns. Even those languages that have a distinct
class of adjectives differ from each other in terms of the properties of adjectives: in some
languages, adjectives “may share at least some of their morpho-syntactic behavior with
nouns, in others they may have more in common with verbs, and still in others they may
Aggrey Wasike. Adjectives in Lubukusu. In Jason Kandybowicz, Travis Major, Harold Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected papers from the
45th Annual Conference on African Linguistics, 325–339. Berlin: Language Science Press.
DOI:10.5281/zenodo.1251746
Aggrey Wasike
be more or less independent of those other classes” (Chafe 2012: 1). Languages that have a
distinct category of adjectives can also differ in terms of adjective inventory: some, such
as English, have a large number of adjectives. In fact adjectives in English constitute an
open class. But other languages have a closed set of adjectives. An example of such a
language is Hausa which has approximately 12 adjectives (Whaley 1997).
As has already been pointed out, some languages lack a distinct category of adjectives.
Instead they express adjectival meanings by use of either verbs or nouns. Examples of
languages which use verbs to express adjectival meanings include Mohawk (Baker 2001),
Seneca and other northern Iroquoian languages (Chafe 2012), Manipuri (Bhat 1994) and
Mayali. To illustrate how a languages use verbs to express adjectival meanings consider
sentences in (1) and (2) from Manipuri and Seneca respectively.
(1)
a. Manipuri (Bhat 1994: 190)
ǝy mabu u-de
I him-acc see-neg(nfut)
‘I did not see him.’
b. phi ǝdu ŋaŋ-de
cloth that red-neg(nfut)
‘That cloth is not red.’
(2)
a. Seneca (Chafe 2012: 10)
Tganӧhso:t,
cis-n.sg.agt -building-upright-sta
t-ka-nӧhs-o:t-ø
‘The house there,’
b. ganӧhsasdë. n.sg.agt -building-big-sta
ka-nöhs-astë-:
‘it was a big house.’
Notice that in these languages words that express adjectival meanings take inflectional
morphology of verbs such as negation in (1).
An example of a language that uses nouns to express adjectival meanings is Quechua.
In this language words that express adjectival meanings take nominal inflectional morphology such as case. This is illustrated in (3).
(3)
a. Peru Quechua (Weber 1983: 6)
rumi-ta rikaa
stone-acc 1sg.see
‘I see a/the stone.’
b. hatun-ta rikaa
big-acc 1sg-see
‘I see a/the big (one).’
Given the cross-linguistic variability illustrated above, it is understandable why linguists would fail to agree on whether the category of adjective is universal or not. Clearly,
326
20 Adjectives in Lubukusu
studying how adjectival meanings are expressed in languages that have not been studied
yet is important as it can improve our understanding of adjectives.
In this paper, I show how adjectival meanings are expressed in Lubukusu. Questions
that I seek to answer include, though not limited to the following: How are adjectival
meanings expressed in Lubukusu? Is Lubukusu English-like, Mohawk-like or Quechualike? How similar or different is Lubukusu from other languages? What nominal and verbal features do words that express adjectival meanings have? Why shouldn’t Lubukusu
adjectives be considered nouns?
I show that Lubukusu has pure adjectives like English, but it not exactly like English
in all respects. This is because there are certain adjectival meanings that are expressed
by use of verbs just like in Mohawk, Seneca, Mayali and Manipuri. Thus Lubukusu has a
mixed adjectival system. I also show that that although adjectives are structurally similar
to nouns, they constitute a lexical category that is distinct. Similarly, although certain
adjectival meanings are expressed by use of verbs, it is cannot be true to argue that
Lubukusu lacks adjectives.
This paper is organized as follows. §2 is a general description of adjectives in Lubukusu
that recognizes two major subtypes of adjectives: basic adjectives and derived adjectives.
§3 describes and discusses nominal features of adjectives focusing on morphological and
syntactic features. §4 is a discussion of verbal features of adjectives. I show that only
stage-level readings of predicates are expressed by use of verbs in Lubukusu. §5 is the
conclusion.
2 Overview of adjectives in Lubukusu
Adjectives are nominal modifiers, and as illustrated in (4) they can occur in a Noun
Phrase (NP) with other modifiers such as numerals, possessive pronouns, demonstratives, associative Prepositional Phrases (PP) and relative clauses.
(4) Lubukusu
ba1 -ba-ana ba-taru ba-nge baa-bofu be lii-ria
baa-kon-a
ba-no
2-22 -person 2-three 2-mine 2.2-big of 5.5-respect 2.rel-sleep-fv 2-dem
‘these three big respectful children of mine who are sleeping’
Because of the constraints of space numerals, possessive pronouns, associative PPs
and relative clauses are not discussed in this paper. I have also not discussed word in
the NP. Instead I have focused on adjectives only, without reference to the other noun
modifiers.
Adjectives in Lubukusu can be divided into two broad categories: basic adjectives and
derived adjectives.
2 Orthographic
B or b is [𝛽] when it is not preceded by a nasal sound
the sequence of these numerals here and in the rest of the paper, the first numeral is the pre-prefix
(augment); the second numeral is class prefix
2 In
327
Aggrey Wasike
2.1 Basic adjectives
These are adjectives that are base generated and are not derived from other lexical categories. Basic adjectives describe size, color, quantity etc. and include -bofu ‘big’, -titi
‘small’, -leyi ‘tall’, -imbi ‘short’, -balayi ‘wide’, -mali ‘black’, -wanga ‘white’, -besemu
‘red’, -kali ‘many’, -lendafu ‘stupid’, -kesi ‘clever’, -silu ‘stupid’, -kara ‘lazy’, -miliu ‘clean’,
-nyalu ‘dirty’, -siro ‘heavy’, -angu ‘light’, -chou ‘big/fully grown’, -khulu ‘old’, -bisi ‘raw/
unripe’, -robe ‘ripe’ etc.
2.2 Derived adjectives
Lubukusu has a very productive process of deriving adjectives from verbs. The process
involves suffixing the root of the verb with the vowel -e3 . In other words the final vowel
of the verb root is replaced with the vowel -e. This suggests that -e is an ‘adjectivizing’ suffix. Derived adjectives ending in -e include-funge ‘closed’ (from funga ‘close’),
-funule ‘open’ (from funula4 ‘open’), -funikhe ‘covered’ (from funikha ‘cover’), -singe
‘washed’ (from singa ‘wash’), -tekhe ‘cooked’ (from tekha ‘cook’), -lekhule ‘free’ (from
lekhula ‘free/let go’), -khalange ‘fried’ (from khalanga ‘fry’), -robe ‘ripe’ (from roba ‘become ripe’), -simbe ‘busy’ (from simba ‘become busy’), -lume ‘hard’ (from luma ‘become
hard’), -sye ‘ground’ (from sya ‘grind’), -ake ‘weeded’ (from yaka ‘weed’), -male ‘smeared’
(from mala ‘smear’), -chichunge ‘sifted’ (from chichunga ‘sift’), -osye ‘roasted’ (from yosya
‘roast’) etc.
The derivation of adjectives from verbs with roots that end in liquids (/l/ & /r/) and
the voiceless fricative (/x/) involve phonological processes that are different from -e final derived adjectives illustrated in the previous paragraph. To form adjectives from
these verbs, the root final liquids and /x/ are changed into fricatives – either [f] or [s]
and the final vowel is changed to [u] or [i]. Adjectives formed in this manner include:
-mesi ‘drunk’ (from mela ‘become drunk’), -changalafu ‘insipid’ (from changalala ‘be
dull, insipid, tasteless), -randafu ‘brown’ (from randara ‘become brown’, -lendafu ‘stupid,
slow’ (from lendala ‘be stupid, be slow thinking), -angafu ‘mature’ (from angala ‘mature’), -nyindafu ‘brave’ (from nyindala ‘brave’), -labufu ‘dirty’ (from labukha5 ‘become
dirty’), rundubafu ‘big’ (from rundubara ‘become big’), -khalafu ‘sad’ (from khalala ‘be
sad’), -nefu ‘fat’ (from nera ‘become fat’), -balakafu ‘dry’ (from balakala ‘dry’), -kafu
‘stupid’ (from kala ‘be stupid/slow thinking’), -khandyafu ‘proud’ (from khandyaba ‘walk
proudly’) etc.
In addition to the two types of derived adjectives exemplified above, there are derived adjectives that end in the suffix -a. These adjectives are fewer compared to the first
two sub-types as they are restricted to occurring with only few nouns. Derived adjectives that end in -a include -fumba ‘folding’ (from fumba ‘fold’, in endebe efumba ‘fold3
Verbs that end in -e should not be confused with adjectives that are derived from verbs which also end in -e.
The two are distinguishable even without reference to tonal patterns: derived adjectives have an augment
and class prefix but verbs have subject agreement and verbal inflections such as tense.
4
Verbs in Lubukusu in the imperative mood occur without any prefixes
5
Orthographic kh- is [x]
328
20 Adjectives in Lubukusu
ing chair’), -chwisya ‘knotted’ (from chwisya ‘knot’, in kumukoye kumuchwisya ‘knotted rope’), -chunula ‘unknotting’ (from chunula ‘unknot’, in kumukoye kumuchunula
‘unknotting rope’, -khalisya ‘crossing/shortcut’ (from khalisya ‘cross/shortcut’, in engila
ekhalisya ‘short cut’, -kisa ‘hide’ (from kisa ‘hide’, in kumwinyawe kumwikisa ‘hide and
seek game’, -kara ‘dribble’ (from kara ‘dribble’, in kumupira kumukara ‘dribbling ball
game (=soccer)’, -chururusya ‘uninterrupted’ (from chururusya ‘let go uninterrupted’ in
kumupira kumuchururusya ‘game where two people hit the soccer ball back and forth’).
As already pointed out, adjectives in this subcategory are few. This may be due semantic
reasons. For example only few objects can be described as folding6 , only few things can
be described as knotting, only a handful on things can be described as dribbling etc. But
it is also possible that these are noun-verb compounds rather than adjectives.
Up to this point we have seen that adjectives in Lubukusu can be formed from verbs.
But the reverse is also possible: verbs can be formed from adjectives through a fairly productive process of suffixing -a to the root of the adjective. To briefly illustrate, adjective
roots such, -imbi ‘short’, -bofu ‘big’, -leyi ‘tall’, and -besemu ‘red’ can be converted into
the following verbs respectively: imbia ‘become short’, bofua ‘become big’, lea ‘become
tall’ and besema ‘become red’. The suffixation of -a to -imbi ‘short’ and -bofu ‘big’ to form
a verb is clear and cannot be contested. These two examples show that the direction of
conversion is from adjective to verb. I assume that this is also true in both lea ‘become
tall’ and besema’ even though these two examples involve additional phonological processes beyond suffixation of -a. In any case the roots -imbi, -bofu, -leyi, and -besemu are
unequivocally adjectival, and conversion must be from adjective to verb.
To summarize this section, we have seen that Lubukusu has many adjectives, some
of which can be classified as basic and others as derived. It is particularly instructive
that the morphological process which forms adjectives from verbs is quite productive. It
is therefore reasonable to conclude that the class of adjectives in Lubukusu is an open
class. This differs from Bantu languages such as Chichewa which has “…very few ‘pure’
adjective stems…” (Mchombo 2004: 24). It also differs from Kiswahili which lacks a productive process of forming adjectives from verbs. For example Kiswahili lacks the noun
+ verb-derived-adjective equivalent of the Lubukusu enyama endekhe ‘cooked meat’. In
Kiswahili to say cooked meat, one must use a relative clause -nyama iliyopikwa ‘meat
that is cooked’.
Having provided a general description of adjectives in Lubukusu, we are now ready
to tackle remaining issues that can challenge our conclusion that adjectives are indeed
a separate and independent lexical category in Lubukusu. An examination of Lubukusu
adjectives indicates that they do have what can be considered nominal features on the
one hand and verbal features on the other. It is necessary to discuss these features and
in the process explain how and why adjectives are neither nouns nor verbs.
6
This contrasts with the adjective -fumbe ‘folded’ (from fumba ‘fold’) which is not as restricted semantically.
Many things can be described as folded: engubo efumbe ‘folded cloth’, ekaratasi efumbe ‘folded paper’ etc.
329
Aggrey Wasike
3 Nominal features of adjectives in Lubukusu
By nominal features, I mean those features that are generally thought of as belonging
to nouns. But instead of nouns being the bearers or associates of these features, it is
adjectives that are. When adjectives carry many nominal features, it can be difficult to
clearly and uniquely distinguish between nouns and adjectives, and it can be challenging
to argue for the existence of a separate category of adjectives. In this section I examine
and illustrate morphological and syntactic nominal features borne by or associated with
the Lubukusu adjectives, beginning with morphological features.
3.1 Morphology of the adjective
Lubukusu adjectives must agree with the nouns that they modify. For this reason they
have a morphological structure that similar to that of nouns. Like nouns, adjectives have
the structure pre-prefix (augment)-class prefix-root. But it is the noun that determines
the form of the pre-prefix and prefix borne by the adjective. The following data illustrates
the structure of nouns, adjectives and agreement patterns in noun classes 1 to 11.
(5)
a. o-mu-ndu o-mu-bofu
1-1-person 1-1-big
‘big person’
b. ba-ba-ndu baa-bofu
2-2-person 2.2-big
‘big people’
(6)
a. ku-mu-sala ku-mu-bofu
3-3-tree
3-3-big
‘big tree’
b. ki-mi-sala ki-mi-bofu
4-4-tree 4-4-big
‘big trees’
(7)
a. li-li-ino lii-bofu
5-5-tooth 5.5-big
‘big tooth’
b. ka-me-eno ka-ma-bofu
6-6-tooth 6-6-big
‘big teeth’
(8)
a. si-sy-uma sii-bofu
7-7-bead 7.7-big
‘big bead’
330
20 Adjectives in Lubukusu
b. bi-by-uma bii-bofu
8.8-bead 8.8-big
‘big beads’
(9)
(10)
a. e-n-dika e-m-bofu
9-9-bicycle 9-9-big
‘big bicycle’
b. chi-n-dika chi-m-bofu
10-10-bicycle 10-10-big
‘big bicycles’
a. lu-lu-ichi luu-bofu
11-11-river 11.11-big
‘big river’
b. chi-nj-ichi chi-m-bofu
10-10-river 10-10-big
‘big rivers’
As shown in (5–10), the structure of adjective is similar to that of noun in each case
both in terms of the number of morphemes and form of the morphemes. This is due to
agreement requirements: the adjective must agree with the noun they modify. The adjective meets this requirement by copying or reduplicating the prefix form and structure
of the noun it modifies. When a noun’s prefixes are o-mu as in (5a), the adjective must
also have o-mu; when a noun’s prefixes are ku-mu as in (6a), the adjective must also
have ku-mu. The only slight variations in the noun and adjective prefix structure can
be found in (5b), (7a & b), (8a & b), and (9a & b), repeated here as (11), (12), (13) and (14)
respectively.
(11) ba-ba-ndu baa-bofu
1-1-person 1-1-big
‘big people’
(12)
a. li-li-ino lii-bofu
5-5-tooth 5.5-big
‘big tooth’
b. ka-me-eno ka-ma-bofu
6-6-tooth 6-6-big
‘big teeth’
(13)
a. si-sy-uma sii-bofu
7-7-bead 7.7-big
‘big bead’
331
Aggrey Wasike
b. bi-by-uma bii-bofu
8.8-bead 8.8-big
‘big beads’
(14)
a. e-n-dika e-m-bofu
9-9-bicycle 9-9-big
‘big bicycle’
b. chi-n-dika chi-m-bofu
10-10-bicycle 10-10-big
‘big bicycles’
These are not counterexamples to the noun-adjective prefix similarity generalization
since they can be explained phonologically. In (12b) the augment-noun prefix turns up as
ka-me because the phonological process of vowel coalescence has taken place. In (14 &
b), the nasal takes on different forms because it assimilates to the place of articulation of
following stop. And finally in (11), (12a), and (13a & b), prefix haplology has taken place.
Prefix haplology takes when identical prefixes such as ba-ba (class 2), li-li (class 5), si-si
(class 7) and bi-bi (class 8) are followed by nominal root or adjectival root that begins
with a consonant. In (11), ba-ba becomes baa because the root of the adjective begins
with a consonant. This holds for (12) as well where li-li becomes lii because the adjective
root is consonant initial. Prefix haplology is not limited to adjectives alone; it takes place
in nouns as well. This is illustrated in the following data.
(15)
a. baa-soreri (from ba-ba-soreri)
2.2-boy
‘boys’
b. lii-fumbi (from li-li-fumbi)
5.5-cloud
‘cloud’
(16)
a. sii-rekere (from si-si-rekere)
7.7-village
‘village’
b. bii-rekere (from bi-bi-rekere)
8.8-village
‘villages’
For a detailed discussion of prefix haplology and how it is accounted for in phonological theory, see Mutonyi (2000).
I end this section by reiterating the nominal features borne by Lubukusu adjectives.
Lubukusu adjectives bear class prefixes that agree with the noun they modify. In addition, they undergo prefix haplology just like nouns. But is this enough to conclude that
Lubukusu adjectives are a sub-set of nouns rather than an independent lexical category?
332
20 Adjectives in Lubukusu
Before answering this question, let us first examine the ‘noun-ness’ of adjectives at the
level of syntax.
3.2 Syntactic function and position of Lubukusu adjectives
Lubukusu adjectives seem to occupy typical noun positions in the sentence without a
modified noun. The typical noun positions which Lubukusu adjectives can occupy are
subject position, object position and object of preposition position. To illustrate, consider
the following sentences.
(17)
a. O-m-bofu o-yu a-siim-a
o-mw-ana wewe.
1-1-blind dem-1 1-prs-love-fv 1-1-child his/hers
‘That blind person loves his/her child.’
b. Baa-tambi ba-a-sab-ang-a
byaa-khulya.
2.2-poor 2-prs-beg-hab-fv 8.8-food
‘The poor usually beg for food.’
(18)
a. Ku-mu-leeyi kw-a-funiikh-e.
3-3-tall
3-pst-break-fv
‘The tall (one) broke.’
b. Li-li-imbi lya-a-kw-a.
5-5-short 5-pst-fall-fv
‘The short (one) fell.’
(19)
a. Wafula a-a-yet-ang-a
baa-tambi.
1.Wafula 1-prs-help-hab-fv 2.2-poor
‘Wafula usually helps the poor.’
lii.bofu.
b. Mayi a-a-kul-il-e
Mother 1-pst-buy-asp-fv 5.5-big
‘Mother bought the big (one).’
(20)
sii-bofu khu-mesa.
a. Wafula a-a-r-a
Wafula 1-pst-put-fv 7.7-big on.prf-table
‘Wafula put the big (one) on the table.’
b. Wafula a-a-r-a
sii-tabu khu-mu-bofu.
Wafula 1-pst-put-fv 7.7-book 17-17-big
‘Wafula put the book on the big one.’
Thus Lubukusu adjectives can function as subject (17& 18), object (19) and object of
preposition (20). These positions – subject position, object position and object of preposition – are noun positions and there is no doubt nor controversy about it. We are
thus confronted yet again with data and facts that underscore the striking similarity
between nouns and adjectives in Lubukusu. Does this mean that adjectives in Lubukusu
are nouns? This is the question we turn to in the next section.
333
Aggrey Wasike
3.3 Are Lubukusu adjectives nouns?
In section 3.1, we saw that adjectives in Lubukusu have a structure that this similar to
that of nouns. Like nouns they have a pre-prefix and a class prefix. In addition adjectives
undergo prefix haplology just like nouns. And in section 3.2, we saw that Lubukusu
adjectives take typical noun functions of subject, object and object of preposition. These
striking similarities raise the possibility that adjectives are just a sub-type of nouns. If
this is indeed the case, then it will not be justifiable to retain adjective as s separate
lexical category in Lubukusu grammar.
I argue that the nominal features of adjectives that we have seen in previous sections
are not sufficient to make the lexical category of adjective in Lubukusu irrelevant. One
piece of evidence which shows that adjectives and nouns in Lubukusu are indeed separate lexical categories comes from NPs that contain both a noun and adjective. These
NPs show unambiguously that nouns and adjectives are generated in different positions,
suggesting that nouns are not adjectives and vice versa.
Nouns in Lubukusu precede adjectives in the NP and as we have already seen in previous sections, adjectives duplicate the prefix system of the nouns that they modify. Consider word order in the following simple Adjective-Noun structure.
(21)
a.
o-mu-soreri o-mu-leyi
1-1-boy
1-1-tall
‘the/a tall boy’
b. * o-mu-leyi o-mu-soreri
1-1-tall
1-1-boy
(Intended: ‘the/a tall boy’)
c. ku-mu-sala ku-mu-leeyi
3-3-tree
3-3-tall
‘the/a tall tree’
d. * ku-mu-leyi ku-mu-sala
3-3-tall
3-3-tree
(Intended: ‘the/a tall tree’)
Clearly, the noun must precede the adjective in the NP. This is significant because it
confirms that Lubukusu adjectives and nouns are base generated in different syntactic
positions.
Notice once again that in (21) as in previous examples that the adjective duplicates
the prefix structure of the noun: in (21a) the adjective duplicates the noun’s o-mu prefix,
while in (21c), the adjective copies the noun’s ku-mu prefix. This type of agreement is referred to as concordial agreement in the Bantu literature. It is this concordial agreement
that explains the rather surprising distribution facts of Lubukusu adjectives illustrated
in (17–20) where adjectives seemed to function as subject, object and object of preposition. Adjectives in such cases contain enough nominal features of nouns (through the
334
20 Adjectives in Lubukusu
prefixes) and can allow for the dropping or omission of the associated nouns without
affecting grammaticality.
Thus NPs such as those in (17–20) that occur without nouns do indeed have a noun
underlyingly. This observation is supported by the fact that interpretation and comprehension of sentences (17–20) is only possible if one knows or has an idea about the nouns
that the adjectives refer to. In other words, these sentences require an appropriate context: they cannot be uttered out of the blue.
The nouns in the underlying structure in (17–20) pass their nominal informationto the
adjectival pre-prefix and prefix through agreement before they are dropped.
To summarize, I have argued that that there is no compelling reason, and there is
no convincing evidence to support an analysis of Lubukusu adjectives as nouns. It is
true that Lubukusu adjectives do indeed have nominal features but this is not entirely
surprising. Lubukusu just happens to be a language (among many others perhaps) where
adjectives share some features with nouns. This tendency by adjectives to share some
of their morphosyntactic features with nouns has long been recognized in some world
languages (Chafe 2012).
The conclusion of this section, then, is that nouns and adjectives exist in Lubukusu as
separate lexical categories.
4 Verbal features of Lubukusu adjectives
A sub-set of adjectives or more broadly adjectival meanings show a relationship with
verbs in Lubukusu. In particular some adjectival meanings are expressed by use of verbs
rather than true adjectives. In (22) for example, the Lubukusu equivalents for ‘happy’,
‘sad’ and ‘tall’ which are unambiguously adjectives in English, are verbs as evidenced by
the fact they bear subject agreement and tense.
(22)
a. Wafula a-a-sangal-il-e.
Wafula 1-prs-happy-asp-fv
‘Wafula is happy.’
b. Wafula a-a-suluny-e.
Wafula 1-prs-sad-fv
‘Wafula is sad.’
c. Wafula a-a-le-il-e.
Wafula 1-prs-tall-asp-fv
‘Wafula has become tall.’
Other examples of English adjectives whose equivalents in Lubukusu are verbs include
the following: lua ‘be tired’, chelewa ‘be late’, khalala ‘be sad’, meniukha ‘be shiny’, imbia
‘become short’, bia ‘become bad’ etc.
In general, the adjectival meanings that are expressed by use of verbs in Lubukusu are
stage-level. These are either ‘non-permanent’, temporary states or continuing processes
335
Aggrey Wasike
or states that are yet to reach their final state. For example sangala ‘be happy’ describes a
temporary, transient state (in contrast to having a happy personality which is expressed
by an adjective as will be illustrated below). Thus to say Wafula aasangalile ‘Wafula
is happy’ means Wafula is happy now, but it doesn’t mean that he will necessarily be
happy later today or tomorrow. Similarly, lea ‘be tall’ does not designate a final state. It
describes a process in progress. Thus to say Wafula aaleile ‘Wafula has grown tall’ means
Wafula has grown taller from last time you saw him, and it doesn’t suggest that he is
done growing. In contrast, expressing the fact that Wafula is a tall person (as his final
tall state) is an individual-level predicate. This in Lubukusu is expressed by an adjective
as will be shown below.
Adjectival verbs such as sangala ‘be happy’ and lea ‘become tall’ occur only in predicative structures, and therefore they are translated in English as predicative adjectives.
Notice that the equivalent English predicative adjectives are obligatorily preceded by BE
in declarative sentences as well as in imperatives.
As already pointed out temporary states and on-going processes adjectival meanings
in Lubukusu are expressed by use of verbs, but permanent final-state attributive adjectival meanings are expressed by use of adjectives. Adjectival meanings of this later type
describe qualities of nouns that are enduring; qualities that are non-temporary. Where
temporary states and on-going adjectival meanings have corresponding permanent attributive meanings, these are expressed by use of adjectives. To illustrate consider (23)
where temporary states and their corresponding permanent states are provided.
(23)
a. Wafula o-mu-sangafu
Wafula 1-1-happy
‘happy Wafula’ (Individual-level)
b. Wafula a-a-sangal-il-e.
Wafula 1-prs-happy-asp-fv
‘Wafula is happy.’ (Stage-level)
c. Ku-mu-sala ku-mu-leyi
3-3-tree
3-3-tall
‘tall tree’ (Individual level)
d. Ku-mu-sala kw-a-le-il-e.
3-3-tree
sa-prs-tall-asp-fv
‘The tree has become tall.’ (Stage-level)
In (23a), there is some permanence to Wafula’s happiness state. Here Wafula has a
happy predisposition; he is naturally a happy person. In (23c) kumuleyi ‘tall’ is an attribute of kumusala ‘tree’: the tree has the attribute tall; it is an attribute that is not
expected to change any time soon. Dixon is therefore correct when he observes that
“…if a language has verbs derived from adjectives, then the adjective is preferred for describing a fairly permanent property and the verb for referring to a more transient state”
(Dixon 2004: 32).
336
20 Adjectives in Lubukusu
Notice that adjectives that express the attributive adjectival meanings (i.e. the true
adjectives) in Lubukusu can be used predicatively. As illustrated in (24), when used predicatively, they retain their adjective forms and do not become verbs.
(24)
a. Wafula a-li o-mu-sangafu.
Wafula 1-be 1-1-happy
‘Wafula is a happy person.’
b. Ku-mu-sala ku-li ku-mu-leyi.
3-3-tree
3-be 3-3-tall
‘The tree is tall.’
Thus attributive adjectives can be used as predicatively just like English. The most significant difference between English and Lubukusu (from the point of view of adjectives)
is that Lubukusu (but not English) makes a distinction between the way it expresses temporary adjectival states or meanings on the one hand and permanent attributive qualities.
The temporary states and on-going process adjectival meanings are expressed by verbs
in Lubukusu, but they are expressed by adjectives in English.
A question that arises is whether verbs that express temporary and on-going processes
are adjectives at some level or not. A straight forward way of determining this is showing that adjectival verbs differ in some significant way from regular verbs. This has been
shown to be true in Mohawk and Vaeakau-Taumako. In these languages, verbs that are
used to express adjectival meanings are different from regular verbs. For example Mohawk verbal adjectives contrast with regular verbs in not taking certain aspectual markers and future tense7 (Baker 2001). In Austronesian languages such as Vaeakau-Taumako,
verbal adjectives differ from regular verbs in their ability to occur without tense-aspectmood marking (Naess & Hovdaugen 2011).
It is therefore reasonable to argue that adjectival verbs are adjectives in Mohawk
and Vaeakau-Taumako at some level because they differ significantly from regular nonadjectival verbs.
In Lubukusu, there is no compelling reason to make a similar argument. This is because
adjectival verbs that describe temporary non-final states and continuing processes are
not different from regular verbs in terms of tense-aspect-mood marking. To illustrate
consider the tense-aspect marking on sangala ‘be happy’ in the following data.
(25)
7
a. Wafula a-la-saangal-a
Wafula 1-fut-happy-fv
Wafula will be happy today.
b. Wafula a-kha-saangal-e
Wafula 1-fut-happy-fv
‘Wafula will be happy tomorrow/next week.’
But see Chafe (2012) who found nothing significant that distinguishes ‘adjectival verbs’ from regular verbs
in Seneca, a language that is related to Mohawk
337
Aggrey Wasike
c. Wafula a-li-saangal-a
Wafula 1-fut-happy-fv
‘Wafula will be happy sometime in the remote future.’
d. Wafula a-a-saangal-a
Wafula 1-pst-happy-fv
‘Wafula was happy a long time ago.’
Clearly, sangala can occur with any tense and aspectual marker just like any regular
verb. I take this to be evidence that Lubukusu adjectival verbs that describe temporary
states and continuing processes are verbs and nothing more. They express adjectival
meanings, but they are verbs in the true sense of the word.
What this means is that the Lubukusu adjective system is different from that of Mohawk and Vaeakau-Taumako in spite of the apparent similarities. Both languages and
Lubukusu express some adjectival meanings by use of verbs. But while all verbal adjectives in Mohawk and Vaeakau-Taumako can be argued to be adjectives, the Lubukusu
ones are not: they are true verbs.
5 Conclusion
In spite of the fact that words which express adjectival meanings in Lubukusu have nominal features on the one hand and verbal features on the other, there is enough strong
evidence that support the existence of adjective as a distinct lexical category. Lubukusu
adjectives have a prefix system that is identical to that of nouns, and the adjectives seem
to function as subject and object, but this doesn’t make them nouns. They remain adjectives and they acquire these features and functions by virtue of being modifiers of nouns.
With regards to the adjectival meanings that are expressed by use of verbs, I showed that
only stage-level predicate readings are expressed by use of verbs in Lubukusu. Individuallevel predicate readings are expressed by use of adjectives. I also argued that ‘adjectival
verbs’ in Lubukusu are real verbs. For this reason, Lubukusu is different from Mohawk
and Vaeakau-Taumako where adjectival verbs have been argued to be adjectives at some
level in the grammar. With regards to the existence of the lexical category of adjective,
Lubukusu is like English (but unlike Mohawk and Vaeakau-Taumako): it has a distinct
lexical category of adjective in its grammar. But this is not to suggest that Lubukusu
and English have identical adjective systems. There are significant differences, one of
which is that stage-level predicate readings in Lubukusu are expressed by use verbs, but
in English it is adjectives that are used.
Acknowledgements
I would like to sincerely thank two anonymous reviewers for their useful detailed comments and suggestions. All errors and shortcomings are my own.
338
20 Adjectives in Lubukusu
Abbreviations
Unless indicating person, numbers in glosses indicate noun class prefixes and pre-prefixes.
Abbreviations follow Leipzig glossing conventions, with the following exceptions:
asp
fv
hab
aspect
final vowel
habitual
sa
sta
subject agreement
stative
References
Baker, Mark C. 2001. The atoms of language. New York: Basic Books.
Baker, Mark C. 2003. Lexical categories: Verbs, nouns, and adjectives. New York: Cambridge University Press.
Bhat, D. N. S. 1994. The adjectival category. Studies in Language Companion Series 24.
Chafe, Wallace L. 2012. Are adjectives universal? The case of Northern iroquoian. Linguistic Typology 16(1). 1–39.
Dixon, R. M. W. 1982. Where have all the adjectives gone and other essays in semantics.
Berlin: Mouton.
Dixon, R. M. W. 2004. Adjective classes in typological perspective. In R. M. W. Dixon
& Alexandra Y. Aikenvald (eds.), Adjective classes: A cross linguistic typology, 1–49.
Oxford: Oxford University Press.
Dixon, R. M. W. & Alexandra Y. Aikhenvald (eds.). 2004. Adjective classes: A crosslinguistic typology. Oxford: Oxford University Press.
Mchombo, Sam. 2004. The syntax of Chichewa. Cambridge: Cambridge University Press.
Mutonyi, Nasiombe. 2000. Aspects of Bukusu morphology and phonology. Columbus: Ohio
State University Doctoral dissertation.
Naess, Åshild & Even Hovdaugen. 2011. A grammar of Vaeakau-Taumako. Berlin/Boston:
Walter de Gruyter.
Weber, David J. 1983. Relativization and nominalized clauses in Huallaga Quechua. Berkeley: University of California Press.
Whaley, Lindsay J. 1997. Introduction to typology: The unity and diversity of language.
Thousand Oaks, California: Sage Publications, Inc.
339
Chapter 21
Optional ergativity and information
structure in Beria
Andrew Wolfe
SIL International
Tajeldin Abdalla Adam
Ergativity in Africa is rare; König (2008: 95–96) lists only twelve African languages that
have been described as exhibiting ergative phenomena. Even more rarely does optional
ergative marking (OEM) appear, in which the use of an ergative marker may depend on
information-structural or discourse-pragmatic considerations. McGregor (2010: 1631) lists a
sole instance in Africa.
Previous literature on Beria, a Saharan language, describes a system of focus marking that
shows ergative alignment, wherein one marker, =gu, focuses transitive actors (A) and another, =di, focuses either transitive patients (P) or intransitive single arguments (S) (Jakobi
& Crass 2004: 151–154; Jakobi 2006).
Based on new data from texts and judicious elicitation, we suggest that this =gu functions
as an optional ergative marker which speakers employ in diverse pragmatic and syntactic contexts, not simply for assigning argument focus to the A term (Lambrecht 1994) but
also for identifying an A when the P term is in focus; for highlighting brand new A participants (Prince 1981) in sentence focus contexts; for disambiguating grammatical roles; and
for marking the A of quotative constructions and embedded relative clauses.
We reanalyze =di as a specificational copula (Mikkelsen 2005), contrasting with the predicational copula =i. Speakers may use either of these copulas in cleft constructions in order
to focus constituents other than transitive actors.
1 Introduction
1.1 Optional ergative marking in Africa
The term ergative describes any linguistic system in which the single or principal argument of an intransitive verb (S) patterns with the patient (P) of a transitive verb rather
than with the actor (A) of the transitive verb. See Figure 1.
Andrew Wolfe & Tajeldin Abdalla Adam. Optional ergativity and information structure in Beria. In Jason Kandybowicz, Travis Major, Harold Torrence & Philip T. Duncan (eds.), African
linguistics on the prairie: Selected papers from the 45th Annual Conference on African Linguistics,
341–358. Berlin: Language Science Press. DOI:10.5281/zenodo.1251748
Andrew Wolfe & Tajeldin Abdalla Adam
Nominative
Accusative
P
A
S
Ergative
transitive
intransitive
Absolutive
P
A
S
Figure 1: Ergative-absolutive vs. nominative-accusative (Dixon 1994; Payne
1997: 129–168)
In optional ergative marking (or OEM; McGregor 2009; 2010; McGregor & Verstraete 2010), use or non-use of the ergative marker is conditioned by principles of information structure, discourse, or semantics. In the analysis below we follow the definition
in McGregor (2009: 493):
Optional ergative case-marking refers to the situation in which the ergative marker
may be present or absent from the Agent NP without affecting the grammaticality
or interpretation of the clause in terms of who is doing what to who. The term
‘optional case-marking’, like ‘free variation’, is potentially misleading, and […] does
not mean that the marker is used randomly.
Only rarely does ergativity appear in descriptions of African languages. König (2008:
95–96) lists twelve languages with potential ergative phenomena, clustered in four families: West Nilotic, Kordofanian, Mande, and Chadic. McGregor (2010: 1631) offers only
one of the West Nilotic languages from König’s list, Shilluk, as the lone example of OEM
from the over 2,000 languages in Africa.
1.2 Beria
Using new data from texts and elicitation, we propose adding Beria, a Saharan language
of eastern Chad and western Darfur, to this short list of African languages known to
exhibit OEM.
In Jakobi & Crass’s (2004) grammar of the Kube dialect of Beria (henceforth J&C, 2004:
151–154; cf. Jakobi 2006), they present two enclitic focus markers, =gu and =di. They argue
that these two focus markers follow an ergative distribution pattern: speakers employ
=gu to mark focused A terms and =di to mark focused P and S terms. A third focusmarking strategy they characterize as a cleft construction, as it uses the copula =i. All
three of these, =gu, =di and =i, encliticize to lexical NPs or free pronominals, though most
participant reference tracking in natural discourse with unmarked pragmatics appears
only in Beria’s verbal cross-reference system, wherein suffixes indicate the person and
number of A (agent / actor) arguments and prefixes indicate the person and number of
P (patient / object) arguments.
The present study seeks to extend and refine Jakobi & Crass’s analysis of =gu, =di and
=i by examining contextualized discourse data with the tools of information structure
342
21 Optional ergativity and information structure in Beria
(especially Prince 1981 and Lambrecht 1994). In a preliminary review of two texts from
a local news broadcast genre, we describe a previously unrecognized usage of =gu in
contexts with sentence focus (Lambrecht 1994: 233–235), i.e., where the entire sentence
consists of new information. Specifically, we propose that in our data, this optional use
of =gu only appears with brand new transitive-agent referents (Prince 1981: 233–237). We
also confirm that =gu may also occur in contexts with argument focus (Lambrecht 1994:
228–233) on A.
In addition to cases where the A term features more prominently in its context, we
delineate three specific constructions that feature a less prominent A argument, also
marked with =gu. All of these occur in contexts where the speaker accords higher prominence to the P argument. This prominence on the P may be pragmatic, as with contexts
where P is in argument focus or serves as the subject of a pseudo-passive, or it may be
syntactic, as the head of a transitive relative clause. We also identify a quotative use of
=gu.
Turning from =gu, we reanalyze =di and =i as two different copulas with distinct
semantics, and recast each of their argument-focus usages as variations on the same cleft
construction. Copular =i, which J&C consider a “copula of identification,” we rebrand as
a predicational copula (Mikkelsen 2005), and we call =di a specificational copula. In one
documented difference between the two cleft constructions, OEM marker =gu appears
on the A term in =di clefts but not in =i clefts. We suggest that this difference owes to
the higher transitivity of the =di cleft, where the clause satisfies Hopper & Thompson’s
(1980: 252) high-transitivity criterion of the individuation of O.
2 Overview of key concepts and text sources
2.1 Information structure
Space considerations preclude an extensive review of the literature on information structure and related subdisciplines, where terminological choices and definitions vary widely.
For present purposes it is helpful to highlight Lambrecht’s (1994) definitions of focus as
well as Prince’s (1981) taxonomy of new information, both of which figure into the analysis below.
Lambrecht (1994: 211) offers a shorthand definition of focus as that which is “unpredictable” or “non-recoverable” in any utterance. As a corollary, every sentence has some
element in focus. He delineates three domains of focus: predicate focus, argument focus,
and sentence focus.
The unmarked subject-predicate (topic-comment) sentence type […], in which the
predicate is in the focus and in which the subject (plus any other topical elements)
is in the presupposition, will be said to have predicate-focus structure; the identificational type […] in which the focus identifies the missing argument in a presupposed open proposition will be said to have argument-focus structure; and
the event-reporting or presentational sentence type, in which the focus extends
343
Andrew Wolfe & Tajeldin Abdalla Adam
over both the subject and the predicate (minus any topical non-subject elements),
will be said to have sentence-focus structure. (Lambrecht 1994: 222).
Prince (1981) sketches out a taxonomy of discourse referents in terms of newness and
givenness. New discourse referents may be either brand new, where the speaker asks
the hearer to create a new entity in her mental representation; or unused, where the
speaker invokes a referent new to the discourse but assumed to be familiar to the hearer.
Given discourse referents may be textually evoked, referred to previously in the discourse itself; or situationally evoked, which Prince primarily uses for the discourse
participants themselves (first and second person). Between these new and given categories lie inferable referents, whose existence the hearer is able to deduce by logical or
cultural implication, as the presence of a “bus” in the discourse implies the existence of
a “driver.”
2.2 Text sources
New data analyzed below come from publicly available Radio Dabanga newscasts from
March 18, 2013 and May 1, 2013, downloaded from the Radio Dabanga website (http:
//www.radiodabanga.org) the day following their broadcast. Based in the Netherlands,
Radio Dabanga (henceforth RD) broadcasts daily Darfur-related news in Arabic and various languages of Darfur. The news anchor for the Beria portion of the broadcast, Tayiba
Abdelkarim Abdul, is a well-respected native speaker of the Kube dialect. For the purposes of the broadcast, she translates from Arabic. Having been given the text well in
advance of her studio time, she considers beforehand how best to render the meaning
of each news piece as a whole into Beria. The final recording undergoes a light editing
process before broadcast.
In general, the paragraph-by-paragraph translation method and the linguistic reputation of the translator notwithstanding, some interference from the source language
cannot be ruled out. For present purposes, these concerns are mitigated by two factors:
Arabic – even Darfuri Arabic – has no ergative case marking; and the second author of
this study, as a native Kube Beria speaker, can critically evaluate the naturalness of the
data in what concerns the conclusions below. Nonetheless, future studies will need to
confirm the present findings with a more extensive data set that has no foreign-language
origin. Accordingly, we qualify our conclusions below as preliminary while we work to
assemble a larger, more diverse corpus.
Elicited data from the present study’s second author appear marked with “TAA” below.
In our discussions, elicitation never involved direct translation from English. Rather, we
would discuss a pragmatic communicative context and ask what a native Beria speaker
might say. Additionally, while recognizing the limitations of ungrammaticality judgments as linguistic evidence, in limited contexts below we deemed evaluations of infelicity important enough to the argument to include them.
344
21 Optional ergativity and information structure in Beria
3 Uses of =gu
3.1 A-argument focus constructions with =gu
Based on data like that given in (1)1 , J&C argue that =gu and =di are focus markers
with complementary distribution, whose usage follows an ergative-absolutive alignment
(2004: 151–154):
(1)
a. (Focus markers with ergative alignment Jakobi 2006)
bágʊ=ɔgɔ=gʊ
Ø:kú-gú-r̩-í.
focused A: =gu
wife-poss.3sg=focerg oj:3:pfv:3-call-sj:3-pfv
‘It’s his wife who called him.’ (Jakobi 2006: 136)
focused P: =di
nɛ-gɛr-g-ɪ.
b. náá=dɪ
pp:2sg=focabs oj:2-look.for-sj:1sg-ipv
‘It’s you I’m looking for.’ (Jakobi 2006: 137)
focused S: =di
c. sʊltǎn=dɪ
Ø:nɪ-Ø-ɪ.
sultan=focabs oj:3:die-sj:3-pfv
‘It’s the sultan who has died.’ (Jakobi 2006: 137)
The enclitic =gu only appears with transitive actors. When speakers want to focus patients or intransitive arguments, they invariably cliticize =di or =i to the focused noun
phrase. In light of Beria’s split-S verbal system (Mithun 1991; Dixon 1994: Chapter 4; see
Jakobi 2011 for detailed analysis of the Beria facts), it bears noting that it is ungrammatical to use =gu with the S of even active intransitive verbs such as ‘run’ or ‘leave,’ which
Beria codes with the A suffixes:2
(2) *Ai=gu suk=tu
hiri-g-í.
1s=erg market=dat run-1sg.a-pfv
‘It was I who ran to the market.’ (TAA)
cf. Ai-di suk-tu hirigí.
(3)
*Ai=gu sur-g-í.
1s=erg exit-1sg.a-pfv
‘It was I who left.’ (TAA)
cf. Ai-di surgí.
1 We
have reproduced J&C’s examples as-is, with translation when in French. Note that they wrote before a
working orthography was in place, so their examples use IPA transcription. Abbreviations from their work
that do not follow the Leipzig Glossing Rules are listed in the Abbreviations section.
2
New Beria examples are written in the working orthography approved by local administration (Faris 2006).
High, [+ATR] vowels are represented with a circumflex diacritic: û, î. Mid, [+ATR] vowels remain unmarked,
predictable from the presence of [+ATR] high vowels within the harmonic domain. Lexical tone is not
written. Pluralization of both nouns and verbs uses the simplified orthographic convention of doubling the
final letter of the word, capturing what is in actuality a complex system of tonal patterns. Imperfective
verbs end in –ì and perfective verbs in –í, reducing grammatical tone phenomena to an iconic visual form.
Both rhotics are written r. Note that we have maintained a single representation for enclitics like =gu even
though they harmonize for ATR with their host.
345
Andrew Wolfe & Tajeldin Abdalla Adam
Because the label of ergative properly applies at the level of the particular grammatical construction, not of whole languages (Croft 2001: 132–171), it is unsurprising to find
languages like Beria with both split-S and ergative alignments in operation in different
components of the grammar. Note also that since =gu does not appear with S arguments,
it indeed follows an ergative pattern and not a marked nominative one (König 2008: 138–
203).
In fact, not even all clauses with two core arguments allow =gu on the more agentive argument. Certain low-transitive (Hopper & Thompson 1980) bivalent verbs such as
‘have,’ ‘learn’ and ‘know’ do not accept =gu but may use =di instead to focus the more
agentive argument:
(4)
*Ber=gu arabie kidí.
A arab tigo.
3s=erg car 3.a:have:pfv:sg 1s car have:1s.A:neg
‘It is he who has a car. I don’t have a car.’ (TAA)
cf. Ber-di arabie kidí. Ai arabie tigo.
(5)
ɟàmâl=dɪ
àrmá Ø:áwáá-r̩-ɪ.
Jamal=focabs Arabic oj:3:learn-sj:3-pfv
‘It’s Jamal who has learnt Arabic.’ (Jakobi 2006: 139)
(6) *Hawa=gu tir=egî.
ege=giní.
Maha ege=gino.
Hawa=erg name=1sg.poss know=3a.pfv:aux Maha know=3a.pfv:aux:neg
‘It was Hawa who learned (lit., ‘knew’) my name. Maha didn’t learn it.’ (TAA)
cf. Hawa-di tir-egî ege-giní. Maha ege-gino.
Turning to information structure, among Lambrecht’s three focus categories (see §2.1),
J&C have already established the use of =gu for argument focus, with focus on the A
term.3 Within this usage, argument focus on the A can accomplish at least two distinct
purposes: asserting the identity of the A term, or asserting its role. This becomes clear in
negative assertions as in (7-8). In (7), the identity of the agent is corrected, whereas in
(8) it is the clause’s role assignments that are contradicted. Both contexts use =gu.
(7)
Bur=do=gu kana
tene sai=gi-n-o.
O
gie=i.
boy=that=erg neg.pfv girl hit=3.a.pfv-aux-neg person other=coppr
‘It wasn’t that boy that hit the girl. It was someone else.’ (TAA)
(8)
Bur=gu kana
tene sai=gi-n-o=ru,
tene=gu.
boy=erg neg.pfv girl hit=3.a.pfv-aux-neg=conj girl=erg
‘It wasn’t that the boy that hit the girl. The girl hit the boy.’ (TAA)
3
They do not use Lambrecht’s terms, but argument focus is clearly indicated in their work by cleft translations (Jakobi & Crass 2004: 151–153, examples 232–236 and 248; and in Jakobi 2006: 136, example 19). In
the other four examples given of =gu (2004: 152, example 237; 2006: 137–138, examples 20, 21, and 26), it is
unclear, based on the example and translation given, what the nature of the “focus” is. Some of these are
treated below.
346
21 Optional ergativity and information structure in Beria
3.2 Sentence focus, brand new A construction
Not all uses of =gu entail argument focus, however. In both RD texts, the speaker uses
=gu in contexts such as (9) where the remainder of the sentence is being asserted as new,
not presupposed.
(9) barûgûî
yom_al_ahad=tu, oo
toûra kuni=gu,
arabie
A.few.days.ago Sunday=on
person:pl armed indf.pl=erg vehicle
tijari
ni genî Gireda=re ji-e,
je
Jokhana=ru
commercial indf village Gireda=abl coploc .3-cvb prog Jokhana=dat
si-r-ì=gi,
toûû ki-si-n-e=ge,
bodo ajas=te
go.ipfv.3-3a-ipfv=rel gun:pl 3.a.pfv-shoot-3.a-cvb=rsn that reason=abl
oo
wetti ku-nu-e,
ere oo
sogodî gie noko=ru
person:pl three 3a.pfv-kill:p.pl-cvb again person:pl ten
and one=dat
hiara=gine=ii
giníí.
wounded=3a.pfv:aux:cvb=coppr :pl 3a.pfv:say: pl
‘Last Sunday some armed people(=erg) opened fire on a commercial vehicle that
was from the town of Gireda, going to Jokhana. Because of that they killed three
people, and 11 people have been injured, they said.’ [RD, 3/18/14, 5.1]
The news anchor does not expect the hearer to know there were people who opened
fire on a commercial vehicle on the road from Gireda to Jokhana, nor does she merely
assert that it was certain unnamed armed men who committed this deed; this is a news
broadcast, and the entire utterance is unpredictable.
In the immediately following sentence, she restates and elaborates on the news just
announced.
(10) oo
Ø-nuíí=gi
araa=gi ni Radio Dabanga kerigi er
person:pl 3.p-die:p.pl:pfv=these person:pl indf Radio Dabanga within rel.pr
k-î-í=gi
oo
toûra malisha hakuma=ru
tabi=gi
3.a.pfv-say-pfv=rel person:pl armed militia government=dat affiliated=rel
dîî=ra
hirdee=ra
k-ori-e=gine=re
camel:pl=conj horse:pl=conj 3.a.pfv-mount-cvb=3a.pfv:aux:cvb=nfin
arabie tijari
yom_al_ahad tûî=gi
gardi genî Jokhana=re
vehicle commercial Sunday
evening=this road village Jokhana=abl
ji-e=re
je
Gireda=r ke-si-r-ì,
toûû
coploc .3-cvb=nfin prog Gireda=dat ven-go.ipfv.3-3a-ipfv gun:pl
ki-si-n-e=ge,
bodo ajas=te
oo
wetti
3.a.pfv-shoot-3.a-cvb=rsn that reason=abl person:pl three
ku-nu-e,
ere oo
sogodî gie noko=ru
3a.pfv-kill:p.pl-cvb again person:pl ten
and one=dat
hiara=giníí
giníí.
wounded=3a.pfv:aux:pl 3a.pfv:say: pl
‘One of the relatives of the deceased told Radio Dabanga that the armed,
347
Andrew Wolfe & Tajeldin Abdalla Adam
government-affiliated militia people mounted on camels and horses and opened
fire on a commercial vehicle on Sunday evening on the road coming from the
village of Jokhana to Gireda. Because of that they killed three people and 11
people were injured, they said.’ [3/18, 5.2]
This example well illustrates the optionality of the ergative marking as defined by
McGregor and cited in §1.1: a “situation in which the ergative marker may be present
or absent from the Agent NP without affecting the grammaticality or interpretation of
the clause in terms of who is doing what to who” (McGregor 2009: 493). On the level
of grammatical relations – and, in fact, even of sentence focus vs. argument focus –
nothing changes between (9) and (10), yet =gu appears in (9) and not in (10). In both,
‘armed people’ (oo toûra) ‘shot’ (toûû kisine) a ‘commercial vehicle’ (arabie tijari) and
‘killed’ (kunue) people. In neither sentence is any one argument in focus. Everything
that changes is irrelevant to the syntax of core arguments: the speaker adds the source
of the information (‘one of the relatives of the deceased’), expands on her description of
the transitive agent (characterizing them as ‘government-affiliated militia’), notes attendant action (transitive: they ‘mounted camels and horses’), and shifts some minor details
(mentioning the ‘road’ and changing the aspect of ‘have been injured’ to ‘were injured’).
No salient grammatical details change, but on a second repetition the once brand
new identity and role of the transitive agent have become textually (Prince 1981: 233–
237). All four of the non-quotative tokens of sentence-focus =gu occur with brand new
transitive-agent participants.
3.3 Constructions with backgrounded A=gu, foregrounded P
Another distinct pragmatically marked construction with =gu appears in one token in
the RD texts, which fits the pattern for neither A-argument focus nor sentence focus
constructions:
(11)
Yara=ra Nimra=ra Dalma=ra
genîa ha Sharq al Jabal hie=ru
villages Mt. Sharq al Jebel direction=loc Yara=conj Nimra=conj Dalma=conj
tabi
toûra malishat hakuma=ru
oo
genîa=kî=ra,
villages=these=conj person:pl armed militia government=dat affiliated
gami=giníí.
begî=gu a-we
this=erg go.3a.pl-cvb ambush=3a.pfv:aux:pl
‘The villages around Mt. Sharq al Jebel – Yara, Nimra, and Dalma – these villages
too, these armed government-affiliated militiamen(=erg) went and ambushed.’
[3/18, 3.1]
Here the A is not new material but textually evoked, as shown by the deictic begî ‘this,
above.’ In context this sentence immediately follows the description of another ambush
by these militia, in the ‘villages that lie to the northeast of Kutum’ (line 2.1). The speaker
marks continuity with the previous episode here not only through anaphoric participant
reference (begî) but also with the lexical repetition of gami-giníí ‘they ambushed’ and
348
21 Optional ergativity and information structure in Beria
genîa ‘villages.’ In other words, everything here is textually evoked except the names
of the ambushed villages and the occurrence of a new incident in ‘these villages, too’
genia-kî-ra.
The new village names added to the list of ‘villages ambushed by the Janjaweed’ are
the unrecoverable content in sentence (11). The speaker fronts the brand new P argument and demotes the textually evoked A argument in a sort of pseudo-passive construction. As noted by Zakaria Fadoul Khidir (2005: 80), Beria has no exact equivalent
to the French passive construction because the presence of passive verbal morphology
in Beria requires the total omission of an agentive lexical noun phrase. To express a demoted agent, the verbal morphology must remain the same but the constituent order
changes to PAV and A gets marked with =gu.
Sentence (11) also exemplifies a phenomenon known to the literature as ergative
hopping (Haviland 1979: 155, cited in Rumsey 2010: 1657) whereby an ergative-marked A
term may simultaneously function as the S argument of an intransitive verb (here ‘went
and’) that intervenes between the ergative-marked noun phrase and the bivalent verb
that licenses it. Three tokens of =gu exhibit ergative hopping.
One of J&C’s examples also falls into the pattern where P is promoted and constituent
order is reversed:
(12) bɪɛ kí=dî
ábā égí=gú
Ø:sí-é-r̩-î.
house this=focabs father my=focerg oj:3:build-pfv:3-sj:3-pfv
‘It’s this house that my father has built.’ (Jakobi 2006: 138)
Although the situational context for utterance (12) is not given, it seems likely that
the A term here is part of a presupposition ‘my father built a house’ or ‘my father built
something,’ putting the A ‘my father’ in the pragmatic background (i.e., not in the focus
domain) and focusing on the clefted P term ‘this house.’
3.4 Quotative construction
Two separate quotative constructions account for the remaining tokens of =gu in the RD
texts. In one of them, which is used in our data exclusively for direct reported speech, no
verb of saying appears between the quoted agent and the quotation itself. The speaker
uses =gu to signal to the hearer that she is transitioning from the A argument to a direct
quotation, as in (13):
(13) …genî=gi kerigi oo
jii=gu,
“Ta-rdasin-e
village=this within person:pl coploc .3:pl=erg 1pl.p-come.together-cvb
baa=ru
ere je
korekk=tu ba=gi
je
oû-d-í,”
hand:pl=ins again prog spade:pl=ins mine=this prog dig-1pl.a-pfv
gi-n-e
kîíí.
3.a.pfv-say-cvb 3.a.pfv-say-pfv:pl
‘… the people who are in this village said, “We are coming together and we are
digging this mine out with our hands, and again with spades.”’ [5/1, 4.3]
349
Andrew Wolfe & Tajeldin Abdalla Adam
This “A – direct Quotation – quotative Verb” pattern satisfies Beria’s default APV word
order, with the quotation serving as the P. Quotative OEM constructions are well documented for at least two other languages (Rumsey 2010): Ku Waru (Trans-New Guinea)
and Bunuba (non-Pama-Nyungan).
The existence of a quotative =gu construction sheds light on one of J&C’s examples
as well:
(14)
bágâ
tɔ-gʊ-ɛ
ɟɪɪ
gɪnɛ, bɛr=gʊ tɛbɪ
millet:pl rv-scatter-3:sj:cvb1 lc:3:aff:p sub pp:3=foc1 take:imp
gɪ-n-ɪ.
prf:3-say-3:sj:aff:prf
‘Since the millet had been scattered, he said: “Take it!”’ (Jakobi & Crass 2004: 152)
Without knowing the communicative context we cannot know for sure, but a direct
quotative =gu reading here seems to fit more naturally than an argument focus reading.
The quotative construction appears in information-structurally diverse contexts: In
(13) and the other RD token, the quoted speaker is a brand new referent and the utterance displays sentence focus whereas in (14), the quoted speaker is presumably textually
evoked (by virtue of its pronominal reference) and the utterance thus has predicate focus.
3.5 Subordinate constructions with =gu
In the other quotative construction the quotation is introduced with a relative clause
‘what A said.’ In this relative clause, the A argument appears marked with =gu, as in (15;
quotation omitted for space):
(15)
wakil_amin_al-am
iga
Umam_al_Muttahida hifz_al_salaam
o
person Undersecretary.General mission United.Nations
peacekeeping
kî tir-ogo
Hervé Ladsous gine_îrì, [o=kî=gu
er
gen name-3sg.poss Hervé Ladsous is.called person=this=erg rel.pr
k-î-í]=gi,
[“…”] gine_kîí.
3a.pfv-say-pfv.sg=rel [“…”] he.said
‘The UN Under-Secretary General for Peacekeeping, whose name is Hervé
Ladsous, what this person said: [“”], he said.’ [5/1, 5.1]
In both such tokens in our data, a brand new participant enjoys an elaborated introduction from the speaker, which is topicalized at the beginning of the utterance. The
speaker then proceeds to preface the participant’s quoted speech act with the short relative clause ‘what he said.’ This relative clause uses a reduced noun phrase to refer to
the quoted participant: ‘this person’ o-kî in (16), and in the other example even more
simply ‘he’ ber. Such a reduced form of reference signals that the quoted participant has
become an evoked, less prominent entity; critically it allows the speaker to pivot from
the communicative purpose of introducing the participant to that of telling the hearer
what it was that the participant said.
350
21 Optional ergativity and information structure in Beria
The ergative marking of A constituents embedded within relative clauses extends beyond quotative uses. In (16), tene ‘girl’ functions as both the patient of the verb kidigarí
‘loves’ in the embedded relative clause and as the single argument of the verb karí ‘came’
in the matrix clause. As the pivot, or shared argument between the two clauses, the P
argument enjoys greater syntactic prominence than the A term.
(16) Tene [bur=gu ki-dig-a-r-í]=do
ka-r-í.
girl boy=erg pfv.3-love-pfv.3-3.sg-pfv:sg=that come-3.a-pfv:sg
‘The girl [that the boy loves] came.’ (TAA)
3.6 Summary
Table 1 shows a taxonomy of =gu constructions identified to this point. Two of these
constructions, shown at bottom – the disambiguation of roles shown in ‘it wasn’t that the
boy hit the girl; the girl hit the boy’ and the signaling of transition from quoted speaker to
quoted speech – have no discernible function in assigning prominence, whether information-structural prominence or syntactic prominence. Rather, their only function is to
disambiguate roles and constituents. This in itself provides the strongest argument for
characterizing =gu as primarily a marker of ergativity and not as a marker of focus.
Common to the remaining constructions is the assignment of marked prominence to
the A term of a transitive verb, whether higher prominence, as in the cases of sentence
focus with a brand new A and of argument focus on A, or lower prominence, as in the
cases of A embedded in relative clauses or of PAV word order (due to argument focus on
P or to pseudo-passivization).
Table 1: Summary explanatory matrix of constructions in which A is marked
with =gu
assigning
prominence
higher
prominence of A
main
argument focus on A (§3.1, (7))
sentence focus on brand new A (§3.2)
lower prominence
of A
main (PAV)
clefted P with argument focus (=di), A
in presupposed material (§3.3, (12))
P as subject of pseudo-passive,
evoked, demoted A (§3.3, (11))
relative
disambiguating
P as pivot of clause linkage, A within
relative clause (§3.5)
disambiguating roles (§3.1, (8))
quotative, signaling transition to speech complement (§3.4)
351
Andrew Wolfe & Tajeldin Abdalla Adam
This schema provides a clear account of all data points in our RD texts as well as
elicited data, and may also elucidate examples given in previous literature that did not
seem to express argument focus on the A term. One significant line of evidence remains
unresolved, however: native speaker intuitions as to the function of =gu. The next section
briefly turns to this before moving on to =di in §4.
3.7 Native speaker intuition: =gu as deixis of role
When Zakaria Fadoul Khidir, a native speaker of Kube Beria, discusses =gu under the
category of “passive voice” (2005: 80; see §3.3 above), he glosses it as a “deictic” marker.
While at first glance this does not seem to integrate easily with the analysis above, it
well captures native speakers’ intuitions about =gu, as the second author of this study
attests and as is further confirmed by Amir Libiss (p.c.), a third Beria speaker.
Beria already has two sets of deictic markers, proximal =kî / =gi and distal =to / =do
(Jakobi & Crass 2004: 126), both with wide syntactic distribution. These markers can even
co-occur with =gu, as in the noun phrase o-kî-gu ‘this person=erg’ in (15) above. Clearly
=gu does not encode mere referential deixis.
At the same time, a deictic reading of =gu accords well with the broader social and
cognitive purpose of deixis and demonstrative marking. If, as Diessel (2006: 463) argues,
“demonstratives function to coordinate the interlocutors’ joint focus of attention,” =gu
could be understood not perhaps as pointing at a referential entity per se, but as pointing
at that entity’s grammatical relation within the transitive clause. By using =gu, a speaker
“points” and invites his hearer to focus attention on the agentive grammatical role played
by the indicated referent. Deixis and argument focus perform related social and cognitive
functions in terms of coordinating interlocutors’ attention.
Of course, as argued at length above, =gu appears in more diverse constructions than
just argument focus on the A term. Nevertheless this does not conflict with the singling
out of A-argument focus as the prototypical use of =gu. By way of comparison, if asked
to define the word ‘that,’ most English speakers would likely point – yet it is no less true
that English ‘that’ has also been grammaticalized to serve in discourse-deictic functions
and even as a relativizer. In fact, one would expect the evolution and grammaticalization
of a deictic marker into such varied additional constructions (Diessel 2006, §4.2).
With this foundation laid, we propose deixis of role as a descriptive term for the
prototypical function of optional ergative marking in Beria. This not only incorporates
native speaker insights about how Beria OEM works but it may provide fruitful directions for future OEM theorizing and research.
To fully comprehend the function of Beria’s =gu, it is also necessary to understand
other focus markers in its environment, especially =di, which was previously described
as an absolutive focus marker (Jakobi & Crass 2004: 151–154; Jakobi 2006). §4 turns to
this.
352
21 Optional ergativity and information structure in Beria
4 Copulas and clefts with =di and =i
J&C characterize =di as an absolutive focus marker because of data such as in (1) above.
To assign argument focus to the A term, Beria speakers use =gu but to focus P or S, they
select =di. We reframe =di, however, as a previously unrecognized copula, which can
then be employed in a cleft construction.
The strongest rationale for this shift is that Jakobi recognizes the existence of a “nonverbal predication marker” (2006: 138) =di sharing a presumed “common origin” with
=di the absolutive focus marker. These two are in fact formally identical, as shown in
(17-18, cf. (1b) above):
kese-r-ì=gi
Tayiba Abdelkarim Abdul=di.
(17) O
person speak-3a-ipfv.sg=rel Tayiba Abdelkarim Abdul=copsp .3
‘The person who is speaking is Tayiba Abdelkarim Abdul.’ [5/1, 2.2]
(18) áɪ=dɪ.
pp:1sg=pred
‘It’s me.’ (Jakobi 2006: 138)
Furthermore, in argument focus constructions, =di shares common distribution with
=i. Both =di and =i may focus S and P arguments but not A arguments. On the other
hand, in our data, neither of these markers shows the complex and varied distribution
summarized for =gu in Table 1 above.
As copular verbs, both =di and =i share a negative form, =do, while =gu has no negative
form (cf. 7-8):
(19) ǧǐm=d-ō.
owl=ic-neg
‘It’s not an owl.’ (J&C 2004: 101) (compare 23)
(20) O
arabie kidí=do
ber=d-o
kire=ego=di.
person car 3.a:have:pfv:sg=rel 3s=copsp -neg brother=3sg.poss=copsp
‘The person who has a car is not him; it is his older brother.’ (TAA)
Semantically, a clean division of labor exists between =i and =di. The first classifies or
ascribes an attribute to the topic: X displays the characteristic Y (21-22), or X is a member
of the set Y (23-24):
(21) mɪsā hɛrr=ɪ.
pot full=ic:3:aff
‘The pot is full.’ (J&C 2004: 100)
(22)
bɔɔ=égí
tákkʊrɛ=ɪ.
ram:pl=1s.poss very.fat=ic:3:aff
‘My rams are very fat.’ (J&C 2004: 100)
353
Andrew Wolfe & Tajeldin Abdalla Adam
(23)
gīmm=ī.
owl=ic:3:aff
‘It’s an owl.’ (J&C 2004: 100)
(24) tàmár̩ā=ɪ.
Tama.person=ic:3:aff
‘It’s a Tama.’ (J&C 2004: 100)
The second marker predicates co-extensive reference between the two terms: X is fully
described by Y and there are no other members in the set; see (17-18) above. Accordingly,
following Mikkelsen (2005), we label =i a predicational copula, and =di a specificational
copula. As she explains (2005: 1), predicational copulas “tell us something about the referent of the subject” whereas specificational copulas “says who or what the referent is.”4
These semantics for =di and =i carry over into their corresponding versions of the
cleft construction. In (12) above, for instance, the speaker presupposes ‘a certain house
exists which my father built’ and, through the specificational cleft, he predicates that
‘this is that house.’ The two referents of ‘this house’ and ‘the house that my father built’
are co-extensive. In (25), by contrast, the speaker assumes ‘a certain substance that the
mother gave to the child’ and then through an predicational cleft, clarifies the nature of
that substance. What sort of thing is that which the mother has given the child? It is of
the class of ‘paste.’
(25) gʊʊ=ɪ
ɪà
ɟàá=r
Ø:kɛkk-ɪ.
paste=cop.sg.abs mother child=adv oj:3:give:sj:3-pfv
‘It’s paste that the mother has given to the child.’ (Jakobi 2006: 139)
One complicating factor in this rendering of the facts is the presence of =gu in the =di
cleft in (14) but not in the =i cleft in (25). We propose to resolve this complication with
reference to Hopper & Thompson’s (1980) scalar transitivity criteria. In §3.1, scalar transitivity already helped explain why low-transitive verbs ‘have,’ ‘learn’ and ‘know’ cannot
occur with ergative-marked agents. If =gu correlates with high transitivity, then Hopper
& Thompson’s criterion of the individuation of O (1980: 252–253) may motivate the use
or non-use of =gu in these two sub-constructions of the Beria cleft. The focused element
of a =di cleft is invariably a referential, definite expression like ‘this house’ whereas in
=i clefts it is a non-referential, indefinite class or attribute like ‘paste.’ In other words,
the fronted P or O in a =di cleft is individuated whereas in a =i it is not.5
Distributional, morphological and semantic lines of evidence converge to support the
assertion that =di is a copula. The only counterevidence we are aware of finds natural
explanation in the dynamics of scalar transitivity.
4
Jakobi & Crass (2004: 102–106) catalogue two other Beria copulas: locative/existential ji and comitative bei.
the individuation of O may also be relevant to the quotative =gu construction (§3.4). Rumsey (2010)
theorizes that the tendency of quotative OEM marking to apply more to direct reported speech than to
indirect stems from this same individuation of O criterion, since the shift of voice and perspective involved
in a direct speech act sets it off more sharply from the framing speech act; it is more clearly individuated
than indirect reported speech.
5 In fact,
354
21 Optional ergativity and information structure in Beria
5 Directions for further research
Directions for future research on OEM phenomena in Beria abound. As discussed in §2.2,
the present study, which represents the first results of an ongoing program of Berialanguage corpus collection and discourse analysis, necessarily offers only preliminary
conclusions due to the small corpus size and its foreign-language source material. With
a larger annotated corpus, a higher-resolution picture will emerge of what additional
factors may influence speakers’ choices of when to use =gu, potentially including animacy, discourse macrostructure, activation status, unexpectedness of agency, aktionsart,
and zero anaphora of other core constituents, among others. More robust explorations
of relative clauses, quotative constructions, topic chains, and ergative hopping will complement this fuller analysis.
Another open question is why Jakobi & Crass’s two folk tale texts (2004: 185–192)
include no instances of =gu, even in contexts with transitive predicates and newly introduced participants, such as in (26):
(26)
sàgʊr tɛnɛ tɛbɪ-ɛ-r̩-ɛ
jackal girl take-prf:3-3:sj-cvb1
‘a jackal took a girl, then…’ (Jakobi & Crass 2004: 185)
We suggest that the lack of =gu here owes in part to the hearers’ assumed familiarity
with stock folk tale participants, which among the Beria speech community would perhaps better be classified as unused participants than brand new ones. This is consistent
with the fact that J&C’s storyteller omits, throughout the story, lexical noun phrases
that would clarify the identity of referents for the benefit of uninitiated readers. J&C
supply these identities in square brackets in their French translation – 21 times in the
first four-page text alone. Considerations of genre and register may also be playing a role
here: we expect to see =gu especially in other formal settings calling for precise speech,
such as testimony in traditional courtroom proceedings or recitation of cultural history
unfamiliar to hearers. At least, if anything, the lack of =gu in J&C’s texts would seem to
strengthen the optional ergativity analysis of =gu given above.
The diachronic origins of =gu also remain opaque. Descriptions of case marking for
Western Saharan (WS) languages (surveyed in König 2008: 38–57) demonstrate similar optional case marking to Beria’s: Core arguments go unmarked in default APV constituent order, and the presence of case marking is influenced by “syntactic, semantic
and pragmatic factors” that are “yet to be adequately described” (Hutchison 1986: 193).
While WS’s “nominative” marker, ye in Kanuri-Kanembu and i in Teda-Daza, bears little
formal resemblance to Beria’s =gu, its usage as an optional marker on A terms is not
unlike =gu. In fact, it is not certain that ye / i indeed functions as a nominative marker
uniformly across WS. Hutchison (1986: 203–205) does establish that in some dialects of
Kanuri, ye may mark an S term – though rarely, and only in narrative discourse when
the subject switches – but in Keshirda Dazaga, for instance, Josiah Walters (p.c.) has yet
found no tokens of an S marked with i. Possibly OEM is not uncommon in the Saharan
family.
355
Andrew Wolfe & Tajeldin Abdalla Adam
6 Conclusions
Close scrutiny of =gu data in Beria’s Kube dialect yields much of interest to the typologist, discourse researcher, and Africanist alike. In the analysis of two radio news
broadcasts and a handful of carefully elicited examples, we have teased apart as many as
seven distinct constructions in which =gu marks only the A term of transitive clauses,
and never P or S terms. Speakers employ =gu in these constructions to various communicative ends, including the disambiguation of roles and the assigning of either higher or
lower prominence to a transitive agent. What brings unity to this diversity is the ergative
nature of =gu: It is indeed foremost a case marker and not a focus marker. At the same
time, it is essentially an optional ergative marker, the use of which varies according to
syntactic and discourse-pragmatic principles. This makes it only the second such system
described as such on the African continent.
Furthermore, we offered multiple lines of evidence showing that =di, which has been
described as an absolutive focus marker, is in fact a previously unrecognized copula. We
contrasted this specificational copula =di with the predicational copula =i. Both of these
may be used in a clefted construction to focus an S or P argument. This reconfiguration
may be summarized as in Table 2.
Table 2: Summary of reanalysis of Jakobi 2006
Jakobi & Crass 2004; Jakobi 2006
=gu
=di
focus
focus
=i
focus
focus marker, ergative
focus marker, absolutive
non-verbal predication marker
cleft construction
copula of identification
optional ergative marker
specificational cleft
specificational copula
predicational cleft
predicational copula
In addition, this analysis lends further support to multiple components of Rumsey’s
(2010) account of OEM phenomena in Ku Waru (Trans-New Guinea) and Bunuba (NonPama-Nyungan), especially in its reliance on scalar transitivity, particularly the individuation of O, in the description of a quotative OEM construction, and in the recognition
of ergative hopping phenomena, in a language far removed from Ku Waru and Bunuba
both geographically and genetically.
Finally, drawing from Khidir (2005) and Diessel (2006) we explored reframing Beria’s
OEM as a variety of deixis we called deixis of role, because in its prototypical usage
=gu “points” to the grammatical role of the ergative-marked noun phrase, directing a
listener’s focus and attention to that agent role.
Much work remains to be done to understand Beria information structure as a whole,
beginning with the assembly of a significantly larger and more diverse corpus, but the
present study represents some basic steps forward toward this end. McGregor (2009:
1626) laments that the study of optional case marking suffers from “not enough data, not
356
21 Optional ergativity and information structure in Beria
enough ideas.” It is our hope that the present study makes a modest contribution toward
addressing these lacunae.
Acknowledgements
We gratefully acknowledge insightful responses given after our initial presentation at
ACAL, especially from Colleen Ahland, Michael Ahland and Malte Zimmermann, who
pointed us in helpful directions for both clearer presentation and further literature review and engagement. Thanks also to Ryan Pennington, who tipped us off to a number
of particularly valuable references. Special thanks to Josiah Walters who shared unpublished data from Keshirda Dazaga. We are grateful to Amir Libiss, another Kube speaker
who reviewed our analysis and offered insights. Thanks finally to Deborah Morton for
reviewing an early draft and offering a multitude of incisive comments. All mistakes are,
of course, ours.
Abbreviations
Abbreviations follow the Leipzig Glossing Rules, except for the following:
a
aimpers
adv
aff
conj
coppr
copsp
coploc
ic
ipv
lc
lm
man
med
mid
agent of transitive verb or
agentive single argument of
intransitive verb
impersonal agent
adverbializer
affirmative
conjunctive
predicational copula
specificational copula
locative copula
identificational copula
imperfective
locative copula
lexical morpheme
manner
marker of medium verbs
middle voice
nfin
o
oj
p
non-final verb form
object
object marker
Patient of transitive verb or patientive
single argument of intransitive verb
pp
person pronoun
pred verbless predicate
rel
relativizer
rel.pr relative pronoun
rsn
reason clause
rv
reduced valence
sa
agent-like subject
sj
subject marker
sp
patient-like subject
sub
subordinator
v
verb
ven
venitive
References
Croft, William. 2001. Radical construction grammar. Oxford: Oxford University Press.
Diessel, Holger. 2006. Demonstratives, joint attention, and the emergence of grammar.
Cognitive Linguistics 17(4). 463–489.
357
Andrew Wolfe & Tajeldin Abdalla Adam
Dixon, R. M. W. 1994. Ergativity (Cambridge Studies in Linguistics 69). Cambridge: Cambridge University Press.
Faris, David. 2006. Beria-English English-Beria dictionary. 2nd edn. Iriba, Chad: Association pour le Développement Économique et Social du département de Kobé (ADESK).
Haviland, John Beard. 1979. Guugu Yimidhirr. In R. M. W. Dixon & Barry J. Blake (eds.),
Handbook of Australian languages, 27–180. Canberra: ANU Press.
Hopper, Paul J. & Sandra A. Thompson. 1980. Transitivity in grammar and discourse.
Language 56(2). 251–99.
Hutchison, John P. 1986. Major constituent case marking in Kanuri. Current approaches
to African linguistics 3. 191–208.
Jakobi, Angelika. 2006. Focus in an active/agentive alignment system – the case of Beria
(Saharan). ZAS Papers in Linguistics 46. 129–42.
Jakobi, Angelika. 2011. Split-S in Beria. In Doris Löhr, Eva Rothmaler & Georg
Ziegelmeyer (eds.), Kanuri, Borno and beyond: Current studies on the Lake Chad region (Topics in Interdisciplinary African Studies 22), 91–116. Cologne: Rüdiger Köppe
Verlag.
Jakobi, Angelika & Joachim Crass. 2004. Grammaire du beria (langue saharienne) (NiloSaharan: Linguistic Analyses and Documentation 18). Cologne: Rüdiger Köppe Verlag.
Khidir, Zakaria Fadoul. 2005. Bases et radicaux verbaux: Déverbatifs et déverbaux du beria
(langue saharienne). Köln: Rüdiger Köppe.
König, Christa. 2008. Case in Africa. Oxford: Oxford University Press.
Lambrecht, Knud. 1994. Information structure and sentence form: Topic, focus, and the
mental representation of discourse referents. Cambridge: Cambridge University Press.
McGregor, William B. 2009. Typology of ergativity. Language and Linguistics Compass
3(1). 480–508.
McGregor, William B. 2010. Optional ergative case marking systems in a typologicalsemiotic perspective. Lingua 120. 1610–36.
McGregor, William B. & Jean-Christophe Verstraete. 2010. Optional ergative marking
and its implications for linguistic theory. Lingua 120. 1607–1609.
Mikkelsen, Line. 2005. Copular clauses: Specification, predication and equation. Amsterdam: John Benjamins.
Mithun, Marianne. 1991. Active/agentive case marking. Language 67(3). 510–46.
Payne, Thomas Edward. 1997. Describing morphosyntax. Cambridge: Cambridge University Press.
Prince, Ellen. 1981. Toward a taxonomy of given-new information. In Peter Cole (ed.),
Radical pragmatics, 223–55. New York: Academic Press.
Rumsey, Alan. 2010. ‘optional’ ergativity and the framing of reported speech. Lingua 120.
1652–76.
358
Part V
Semantics and pragmatics
Chapter 22
Two-place exceed comparatives in
Luganda
M. Ryan Bochnak
University of Konstanz
Research on comparative constructions recognizes the need for both a 3-place (“phrasal”)
comparative operator, alongside a 2-place (“clausal”) operator (e.g., Heim 1985; Bhatt & Takahashi 2011). Recent cross-linguistic work on comparatives has argued that exceed comparative constructions are phrasal comparatives, making use of a 3-place operator (e.g., Beck
et al. 2009; Howell 2013 for Yorùbá). While certain exceed constructions in Luganda can indeed be analyzed in this way, I argue here for the idea that others involve a 2-place operator
that compares two degrees directly. I treat nominalized adjectives as measure functions in
the sense of Bartsch & Vennemann 1972 and Kennedy 1997: they map an individual to its
maximal degree on a scale. This allows us to model possessed adjective nominalizations similar to Barker’s (1995) analysis of relational nouns, although whereas for Barker a possessive
DP denotes a predicate of individuals, in this case the resulting DP denotes a degree.
1 Introduction
Formal research on comparatives distinguishes between phrasal comparatives (1) and
clausal comparatives (2), depending on the syntactic category of the standard phrase
(complement of than; angled brackets represent ellipsis).
(1) Kim is taller than [DP Lee].
(2) Kim is taller than [CP Lee is <tall> ].
phrasal
clausal
Along with their different syntax, phrasal and clausal comparatives are taken to have
different semantic representations (Bhatt & Takahashi 2011; Heim 1985; Kennedy 1997:
among others). While phrasal comparatives involve an operator with three argument
positions, clausal comparatives involve an operator with two argument positions (see
(3); the standard semantic analyses of these will be unpacked in §2).
(3)
a. phrasal comparison ↔ 3-place comparative operator
b. clausal comparison ↔ 2-place comparative operator
M. Ryan Bochnak. Two-place exceed comparatives in Luganda. In Jason Kandybowicz, Travis
Major, Harold Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected
papers from the 45th Annual Conference on African Linguistics, 361–375. Berlin: Language Science
Press. DOI:10.5281/zenodo.1251750
M. Ryan Bochnak
In this paper, I challenge the assumption that the syntax-semantics mapping in comparatives is necessarily as in (3), based on a study of exceed-comparison in Luganda
(Bantu; JE15). Recent work has argued that exceed-comparatives are strictly phrasal and
therefore make use of a 3-place comparative operator (Beck et al. 2009; Howell 2013 for
Yorùbá). I too will argue that exceed comparatives in Luganda are syntactically phrasal
comparatives, and that most exceed constructions can be analyzed as involving a 3-place
operator. Exceed constructions in Luganda come in two varieties: the verb okusinga ‘to
exceed’ can appear as the main verb as in (4), or in a subordinate (infinitive) form as in
(5). In main verb exceed constructions, the gradable predicate appears in a nominalized
form (with the noun class 14 prefix bu-), while in subordinate exceed constructions, the
gradable predicate appears as the main predicate of the sentence (showing noun class
agreement with the subject).
(4)
(5)
Kizito asinga
Kato obukulu.
Kizito a-singa
Kato o-bu-kulu
Kizito nc1-exceed Kato aug-nc14-old
‘Kizito is older than Kato.’
lit.: ‘Kizito exceeds Kato in oldness.’
Kizito mukulu okusinga
Kato.
Kizito mu-kulu o-ku-singa
Kato
Kizito nc1-old aug-nc15-exceed Kato
‘Kizito is older than Kato.’
lit.: ‘Kizito is old exceeding Kato.’
main verb exceed
subordinate exceed
However, I will argue that “subcomparatives” in Luganda are syntactically phrasal but
make use of a 2-place “clausal”-like comparative operator. An example of this construction is given in (6), where the two arguments of exceed are delineated by square brackets,
and are both DPs headed by nominalized gradable predicates. (A licit subcomparative in
English is given in the translation line of (6).)
(6) [Obuwanvu bw’
bwayo].
emmeeza]
businga
[obugazi
o-bu-wanvu bu-a
e-N-meeza
bu-singa
o-bu-gazi
bu-ayo
aug-nc14-long nc14-gen aug-nc9-table nc14-exceed aug-nc14-wide nc14-poss
‘The table’s length exceeds its width.’
In order for this idea to go through, I will also need to provide an analysis of nominalized (NC14 bu-marked) adjectives in Luganda. To do this, I will build on the intuitions of
Moltmann (2009) and Nicolas (2004) that nominalizations of gradable adjectives are relational, as well as on the standard degree-based analysis of gradable adjectives (Cresswell
1976; Kennedy & McNally 2005; von Stechow 1984: among others).
The consequences of this analysis are the following: (i) there is novel evidence for
two-place comparatives for exceed languages, which has not previously been adduced;
and (ii) at least some syntactically phrasal comparatives can receive a two-place comparative analysis, contra the mappings in (3). This paper proceeds as follows: §2 provides an
362
22 Two-place exceed comparatives in Luganda
overview of the phrasal vs. clausal distinction; §3 outlines a three-place operator analysis of phrasal comparatives in Luganda; in §4 I argue for the existence of a two-place
operator in Luganda based on evidence from subcomparatives, and I consider two types
of analyses for nominalized gradable predicates; §5 concludes.
2 The composition of comparatives
2.1 Phrasal comparatives
A phrasal comparative like (7) in English can be analyzed as involving the 3-place comparative operator in (8). The arguments of the operator are two individual arguments (the
standard and target of comparison), and the gradable predicate that provides the scale
for comparison. A gradable predicate is taken to denote a relation between an individual
and a degree, as in (9).
(7)
Kim is taller than [DP Lee].
(8)
3-place -er for phrasal comparatives:
J-er3 K = 𝜆𝑥 ∈ 𝐷𝑒 𝜆𝐺 ∈ 𝐷⟨𝑑,⟨𝑒,𝑡⟩⟩ 𝜆𝑦 ∈ 𝐷𝑒 .max(𝜆𝑑.𝐺(𝑑)(𝑦)) ≻ max (𝜆𝑑 ′ .𝐺(𝑑 ′ )(𝑥))
(9) JtallK = 𝜆𝑑𝜆𝑥.height(𝑥) ⪰ 𝑑
(Cresswell 1976; Kennedy & McNally 2005)
A sample derivation of (7) is given in (10)-(11), assuming that DegP is the sister of A′
(Heim 2001), than is semantically vacuous, and the than phrase extraposes at PF.
S
⟨𝑡⟩
(10)
DP
⟨𝑒⟩
VP
⟨𝑒, 𝑡⟩
Kim
V
AP
⟨𝑒, 𝑡⟩
is
A′
⟨𝑑, ⟨𝑒, 𝑡⟩⟩
DegP
⟨⟨𝑑, ⟨𝑒, 𝑡⟩⟩, ⟨𝑒, 𝑡⟩⟩
Deg
⟨𝑒, ⟨⟨𝑑, ⟨𝑒, 𝑡⟩⟩, ⟨𝑒, 𝑡⟩⟩⟩
-er3
A
PP
⟨𝑒⟩
tall
P
DP
than
Lee
363
M. Ryan Bochnak
(11)
JDegK = J-er3 K = 𝜆𝑥⟨𝑒⟩ 𝜆𝐺⟨𝑑,⟨𝑒,𝑡⟩⟩ 𝜆𝑦⟨𝑒⟩ .max(𝜆𝑑.𝐺(𝑑)(𝑦)) ≻ max (𝜆𝑑 ′ .𝐺(𝑑 ′ )(𝑥))
JPPK = 𝑙
JA′ K = JtallK = 𝜆𝑑𝜆𝑥.height(𝑥) ⪰ 𝑑
JDegPK = JDegK(JPPK) = 𝜆𝐺𝜆𝑦.max(𝜆𝑑.𝐺(𝑑)(𝑦)) ≻ max (𝜆𝑑 ′ .𝐺(𝑑 ′ )(𝑙))
JVPK = JAPK = JDegPK(JAPK) = 𝜆𝑦.max(𝜆𝑑.height(𝑦) ⪰ 𝑑) ≻
max(𝜆𝑑 ′ .height(𝑙) ⪰ 𝑑 ′ )
f. JSK=JVPK(JDPK)=1 iff max(𝜆𝑑.height(𝑘) ⪰ 𝑑) ≻ max (𝜆𝑑 ′ .height(𝑙) ⪰ 𝑑 ′ )
a.
b.
c.
d.
e.
The truth conditions for the sentence can be paraphrased as “the maximal degree to
which Kim is tall is greater than the maximal degree to which Lee is tall.”
2.2 Clausal comparatives
Meanwhile, clausal comparatives like (12) are analyzed in terms of the two-place comparative operator in (13). The two arguments of two-place -er are both sets of degrees
(type ⟨𝑑, 𝑡⟩ functions), which are derived in syntax by movement.
(12)
(13)
Kim is taller than [CP Lee is <tall> ].
2-place -er for clausal comparatives:
J-er2 K = 𝜆𝐷1 ∈ 𝐷⟨𝑑,𝑡⟩ 𝜆𝐷2 ∈ 𝐷⟨𝑑,𝑡⟩ .max(𝐷2) ≻ max(𝐷1)
A sample derivation of (12) is given in (14), assuming null operator movement within
than phrase to derive a ⟨𝑑, 𝑡⟩ function (this step not shown here; see Chomsky 1977),
movement of DegP to derive another ⟨𝑑, 𝑡⟩ function, and ellipsis within the than phrase.
364
22 Two-place exceed comparatives in Luganda
(14)
S1
⟨𝑡⟩
DegP1
⟨⟨𝑑, 𝑡⟩, 𝑡⟩
.
⟨𝑑, 𝑡⟩
1
S
⟨𝑡⟩
VP
⟨𝑒, 𝑡⟩
DP
⟨𝑒⟩
Kim
Deg
⟨⟨𝑑, 𝑡⟩, ⟨⟨𝑑, 𝑡⟩, 𝑡⟩⟩
V
-er2
P
than
AP
⟨𝑒, 𝑡⟩
is
PP
⟨𝑑, 𝑡⟩
CP
⟨𝑑, 𝑡⟩
Lee is <𝑑−tall>
DegP
⟨𝑑⟩
A′
⟨𝑑, ⟨𝑒, 𝑡⟩⟩
t1
A
tall
(15)
a.
b.
c.
d.
e.
f.
g.
h.
𝜆𝑑 ′ .height(𝑙)
JPPK =
⪰ 𝑑′
J-er2 K = 𝜆𝐷1 ∈ 𝐷⟨𝑑,𝑡⟩ 𝜆𝐷2 ∈ 𝐷⟨𝑑,𝑡⟩ .max(𝐷2) ≻ max(𝐷1)
JDegPK = J-er2 K(JPPK) = 𝜆𝐷2.max(𝐷2) ≻max(𝜆𝑑 ′ .height(𝑙) ⪰ 𝑑 ′ )
JA′ K = JtallK = 𝜆𝑑𝜆𝑥.height(𝑥) ⪰ 𝑑
JVPK = JAPK = 𝜆𝑥.height(𝑥) ⪰ 𝑑
JSK = height(𝑘) ⪰ 𝑑
J.K = 𝜆𝑑.height(𝑘) ⪰ 𝑑
JS1 K = JDegPK(J.K) = max(𝜆𝑑.height(𝑘) ⪰ 𝑑) ≻max(𝜆𝑑 ′ .height(𝑙) ⪰ 𝑑 ′ )
Note that the exact same truth conditions are derived for 3-place and 2-place comparatives (compare (11f) and (15h)). With this background in place, we now turn to the
analysis of exceed comparatives in Luganda.
3 A 3-place comparative operator in Luganda
Reviewing what we have already seen, comparatives in Luganda are formed using the
verb (oku)singa ‘exceed’, where the direct object of (oku)singa is a DP naming the standard of comparison. Exceed comparatives in Luganda come in two varieties. In (16), the
365
M. Ryan Bochnak
exceed verb is the main verb, with a gradable predicate in nominalized form with the
NC14 bu-prefix. In (17), the exceed verb phrase is in a subordinate (infinitive) form marked
with the NC15 ku-prefix, while the gradable predicate is the main predicate.
(16) Kizito asinga
Kato obukulu.
Kizito a-singa
Kato o-bu-kulu
Kizito nc1-exceed Kato aug-nc14-old
‘Kizito is older than Kato.’
lit.: ‘Kizito exceeds Kato in oldness.’
(17)
Kizito mukulu okusinga
Kato.
Kizito mu-kulu o-ku-singa
Kato
Kizito nc1-old aug-nc15-exceed Kato
‘Kizito is older than Kato.’
lit.: ‘Kizito is old exceeding Kato.’
main verb exceed
subordinate exceed
Comparatives with (oku)singa can be used in quality comparisons (above), as well as
amount comparisons, as shown in (18)-(19).1
(18)
Charlotte yasinga
Rita okuwandiika amabaluwa.
Charlotte a-a-singa
Rita o-ku-wandiika a-ma-baluwa
Charlotte nc1-pst-exceed Rita aug-nc15-write aug-nc6-letter
‘Charlotte wrote more letters than Rita.’
lit.: ‘Charlotte exceeds Rita in writing letters.’
main verb exceed
(19) Charlotte yawandiika amabaluwa
okusinga
Rita.
Charlotte a-a-wandiika a-ma-baluwa o-ku-singa
Rita
Charlotte nc1-pst-write aug-nc6-letter aug-nc15-exceed Rita
‘Charlotte wrote more letters than Rita.’
lit.: ‘Charlotte wrote letters, exceeding Rita.’
subordinate exceed
Given that the object of (oku)singa is a DP, this looks like phrasal comparison. Let
us propose that (oku)singa has the semantics of three-place comparative -er3 as in (20),
where the direct object of (oku)singa (i.e., the standard of comparison) is an individualdenoting DP. Evidence for a degree-based analysis for Luganda comes from the fact that
both the main verb and subordinate exceed constructions pass Kennedy’s (2007a) tests for
explicit comparison, including acceptability in crisp judgment contexts, and the ability to
be formed with absolute-standard gradable predicates (see Bochnak 2013 for these tests).
This proposal thus makes Luganda similar to other exceed languages, such as Yorùbá,
which has also been argued to use a 3-place operator (Beck et al. 2009; Howell 2013).
(20) J(oku)singa3 K = J-er3 K =
𝜆𝑥 ∈ 𝐷𝑒 𝜆𝐺 ∈ 𝐷⟨𝑑,⟨𝑒,𝑡⟩⟩ 𝜆𝑦 ∈ 𝐷𝑒 .max(𝜆𝑑.𝐺(𝑑)(𝑦)) ≻ max (𝜆𝑑 ′ .𝐺(𝑑 ′ )(𝑥))
1I
assume a covert adjective ma-ngi ‘many’ is present in (18)-(19) to derive the amount comparison reading.
The adjective may also appear overtly, but not shown here due to space. See Bochnak 2013.
366
22 Two-place exceed comparatives in Luganda
The proposed structure and derivation of the main verb exceed comparative in (16) is
given in (21)-(22).2
(21) Kizito asinga Kato obukulu.
‘Kizito exceeds Kato in oldness.’
S
VP2
DP
Kizito
(22)
a.
b.
c.
d.
e.
VP1
DP
V
DP
asinga
Kato
obukulu
JasingaK = 𝜆𝑥𝜆𝐺𝜆𝑦.max(𝜆𝑑.𝐺(𝑑)(𝑦)) ≻ max(𝜆𝑑 ′ .𝐺(𝑑 ′ )(𝑥))
JobukuluK = 𝜆𝑑𝜆𝑥.old(𝑥) ⪰ 𝑑
JVP1 K = 𝜆𝐺𝜆𝑦.max(𝜆𝑑.𝐺(𝑑)(𝑦)) ≻ max(𝜆𝑑 ′ .𝐺(𝑑 ′ )(𝐾 𝑎𝑡𝑜))
JVP2 K = 𝜆𝑦.max(𝜆𝑑.old(𝑦) ⪰ 𝑑) ≻ max(𝜆𝑑 ′ .old(𝐾 𝑎𝑡𝑜) ⪰ 𝑑 ′ )
JSK = 1 iff max(𝜆𝑑.old(𝐾 𝑖𝑧𝑖𝑡𝑜) ⪰ 𝑑) ≻ max(𝜆𝑑 ′ .old(𝐾 𝑎𝑡𝑜) ⪰ 𝑑 ′ )
This analysis derives the intuitively correct truth conditions for the comparative: (16)
is true if and only if the maximal degree to which Kizito is old is greater than the maximal degree to which Kato is old. It thus provides a straightforward derivation of main
verb exceed constructions parallel to English phrasal comparatives. However, note that
I have assigned a semantics for nominalized bu-kulu ‘nc14-old’ identical to that of gradable adjectives in English (cf. 9). If the underlying semantics of the gradable adjective
stem -kulu ‘old’ is the same as gradable adjectives in English, this means that the nominalization morphology bu- is semantically vacuous. This assumption will be revised later
on in §4, but for now it allows us to straightforwardly derive the truth conditions for the
comparative with the semantic tools familiar from English.
For subordinate exceed constructions, let us assume the same three-place comparative
operator semantics for (oku)singa in (20). Given the (simplified) structure for (17) as in
(23), the semantic derivation proceeds as in (24).
(23) Kizito mukulu okusinga Kato.
‘Kizito is old, exceeding Kato.’
2
See Bochnak 2013 for arguments for the syntax proposed here.
367
M. Ryan Bochnak
S
DP3
Kizito
AP2
AP1
mukulu
DP2
aug
o-
(24)
a.
b.
c.
d.
e.
NP
VP
kuV
DP1
-singa
Kato
J-singaK = 𝜆𝑥𝜆𝐺𝜆𝑦.max(𝜆𝑑.𝐺(𝑑)(𝑦)) ≻ max(𝜆𝑑 ′ .𝐺(𝑑 ′ )(𝑥))
JDP2 K = 𝜆𝐺𝜆𝑦.max(𝜆𝑑.𝐺(𝑑)(𝑦)) ≻ max(𝜆𝑑 ′ .𝐺(𝑑 ′ )(𝐾 𝑎𝑡𝑜))
JmukuluK = 𝜆𝑑𝜆𝑥.old(𝑥) ⪰ 𝑑
JAP2 K = 𝜆𝑦.max(𝜆𝑑.old(𝑦) ⪰ 𝑑) ≻ max(𝜆𝑑 ′ .old(𝐾 𝑎𝑡𝑜) ⪰ 𝑑 ′ )
JSK = 1 iff max(𝜆𝑑.old(𝐾 𝑖𝑧𝑖𝑡𝑜) ⪰ 𝑑)) ≻ max(𝜆𝑑 ′ .old(𝐾 𝑎𝑡𝑜) ⪰ 𝑑 ′ )
Parallel truth conditions are thus derived for (16) and (17), and indeed these constructions appear to have the same truth conditions.3 Note once again the assumption that
nominalization morphology (this time ku- on the verb phrase) is semantically vacuous.
In sum, both main verb and subordinate exceed comparatives in Luganda receive a
straightforward analysis as three-place phrasal comparatives by borrowing the familiar
tools from English and other better-studied languages. This seems like a nice result, given
that other exceed languages have also been analyzed in this way.4 Furthermore, it has
been noted for other languages that only three-place comparatives exist (e.g., Bhatt &
Takahashi 2011 on Hindi/Urdu). That is, there is no reason to expect that a language
necessarily uses both 3-place and 2-place comparison. However, in the next section I
present evidence for the existence of two-place comparatives in Luganda, and propose
an analysis involving a two-place version of the exceed verb.
3
Again, they are both acceptable in crisp judgment contexts and are productive will gradable predicates of all
scale structures (Bochnak 2013). However, my consultants seem to have a preference for subordinate exceed
constructions, in that they are almost always offered first as translations of English. The corresponding
main verb versions are nevertheless always accepted by speakers when offered by the researcher. I have
no explanation for this apparent preference.
4
In an LFG-based analysis, Beermann et al. (2005) likewise propose a common semantics for comparison
between English and Luganda, despite their different syntactic structures.
368
22 Two-place exceed comparatives in Luganda
4 A 2-place comparative operator in Luganda
One of the tests for diagnosing clausal (2-place) comparatives is the availability of multiple standards of comparison (Lechner 2001; Merchant 2009; Bhatt & Takahashi 2011).
For instance, (25) contains three standards of comparison following than. The idea is this:
the apparent phrasal surface form of comparatives like Kim is taller than Lee is the result
of ellipsis applying to an underlyingly clausal complement of than, where the standard
has moved out of the ellipsis site. Multiple standards can appear after than so long as
they have all moved out of the ellipsis site.5
(25)
Kim read more books on Tuesday than Lee magazines on Thursday.
Multiple standards in Luganda, however, are not licensed, either in main verb (26) or
subordinate (27) exceed constructions.6
(26)
* [Charlotte] [ku mande] yasinga
[Rita] [ku lw’okubiri] okuwandiika
Charlotte ku mande a-a-singa
Rita ku lw’okubiri o-ku-wandiika
Charlotte loc Monday nc1-pst-exceed R
loc Tuesday aug-nc15-write
amabaluwa
amangi.
a-ma-baluwa a-ma-ngi
aug-nc6-letter aug-nc6-many
Intended: ‘Charlotte wrote more letters on Monday than Rita wrote
on Tuesday.’
main verb
(27)
* Charlotte yawandiika amabaluwa
mangi
ku mande okusinga
Charlotte a-a-wandiika a-ma-baluwa ma-ngi ku mande o-ku-singa
Charlotte nc1-pst-write aug-nc6-letter nc6-many loc Monday aug-nc15-exceed
[Rita] [ku lw’okubiri].
Rita ku lw’okubiri
Rita loc Tuesday
Intended: ‘Charlotte wrote more letters on Monday than Rita wrote
on Tuesday.’
subordinate
A second test for the availability of clausal comparatives comes from subcomparatives: comparisons based on two different dimensions. An English example is given in
(28). In this case, the complement of than has overt clausal syntax.
(28)
The table is longer than it is wide.
5 See
Lin (2009) for arguments that comparatives with multiple standards in Chinese are still phrasal in
nature, and not the result of ellipsis from a clausal source. In other words, the ability to have multiple
standards does not necessarily entail that there is an underlying clausal source, so this test does not provide
strong evidence for diagnosing clausal comparatives.
6
(26)-(27) contain the adjective ma-ngi ‘many’; cf. footnote 1 and (18)-(19). Its presence/absence does not
affect the grammaticality of these sentences.
369
M. Ryan Bochnak
Versions of these types of comparisons are possible using the Luganda main verb
exceed construction, as shown in (30). However, the constituents forming the comparison
are not full clauses. Rather, the subject and object of okusinga are headed by nominalized
gradable adjectives (bolded in 30), which name the two dimensions of comparison. These
nominalizations appear in a possessive construction, the general form of which is given
in (29), where gen is the genitive particle.7
(29)
possessed adjective nominalizations = [bu-adj + gen + possessor]
(30) Obuwanvu
bw’
emmeeza
businga
obugazi
bwayo.
o-bu-wanvu bu-a
e-N-meeza
bu-singa
o-bu-gazi
bu-ayo
aug-nc14-long nc14-gen aug-nc9-table nc14-exceed aug-nc14-wide nc14-poss
‘The table’s length exceeds its width.’
I suggest that this is a case of 2-place comparison in Luganda, despite the fact that
these are not syntactically clausal comparatives (i.e., no clausal syntax in the standard).
To see how this would work, we need an analysis of possessed nominalized adjectives. I
consider two styles of analysis here, which both deliver the desired result.
The first style of analysis I will call the relational analysis, following the intuitions of Moltmann (2009) and Nicolas (2004) that nominalized properties like length
are inherently relational. That is, a nominalized adjective like bu-wanvu (‘nc14-long’ ≈
‘length’/‘longness’) relates an individual to its length. Given that gradable adjectives are
also standardly taken to denote relations between individuals and degrees (Cresswell
1976; Heim 2001; Kennedy & McNally 2005), it seems at first blush that the assumption
we made in §3 that nominalization is vacuous is a reasonable one. However, rather than
assuming that nominalizing bu- is vacuous, I propose it has the function of reversing the
argument relations of a gradable adjective; the difference between the gradable adjective
and its nominalized form is thus given in (31)-(32).
(31)
(32)
J-wanvuK = JlongK = 𝜆𝑑𝜆𝑥.length(𝑥) ⪰ 𝑑
Jbu-wanvuK = 𝜆𝑥𝜆𝑑.length(𝑥) ⪰ 𝑑
⟨𝑑, ⟨𝑒, 𝑡⟩⟩
⟨𝑒, ⟨𝑑, 𝑡⟩⟩
This change of argument structure means that a nominalized adjective expects an
individual as its first argument, which will be the possessor 𝑥 of the property. After
saturating the individual argument position, we are left with a set of degrees that 𝑥’s
length is greater than or equal to. Thus, the DP obuwanvu bw’emmeeza (≈ ‘the length of
the table’) denotes the set of degrees in (33).
(33) Jobuwanvu bw’emmeezaK = 𝜆𝑑.length(𝑡) ⪰ 𝑑
For a compositional analysis, I follow the syntax of possession proposed by Barker
(1995). On Barker’s analysis, the possessive DP is headed by a possessive D head poss,
which is null in English. For non-relational nouns (e.g., table), poss introduces the possessive relation 𝜋, as shown in (34). For inherently relational nouns, poss is still present,
7 Note
370
that bu- on gen,poss, and -singa are inflectional prefixes for agreement with a nc14 noun.
22 Two-place exceed comparatives in Luganda
but simply denotes the identity function (𝜆𝑅.𝑅). Barker’s analysis of possessed relational
nouns in English in outlined in (35)-(36).
(34) JpossK = 𝜆𝑃𝜆𝑦𝜆𝑧.𝜋(𝑦, 𝑧) ∧ 𝑃(𝑧)
(35)
DP2
D′
DP1
John’s
(36)
NP
D
child
poss
a. JchildK = 𝜆𝑥𝜆𝑦.child(𝑥, 𝑦)
b. JpossK = 𝜆𝑅.𝑅
c. JJohn’s childK = 𝜆𝑦.child(𝑗, 𝑦)
I propose for Luganda that the genitive particle -a is the overt spell-out of Barker’s
poss. The analysis for the possessive nominalized adjective obuwanvu bw’emmeeza ‘the
length of the table’ is thus given in (37)-(38), whereby the nominalized adjective denotes
a relational noun, and the result of applying the possessor is a (characteristic function
of a) set of degrees of length.
(37)
DP2
D′
(38)
a.
b.
c.
d.
DP1
NP
D
obuwanvu
bu-a
emmeeza
JobuwanvuK = 𝜆𝑥𝜆𝑑.length(𝑥) ⪰ 𝑑
J-aK = JpossK = 𝜆𝑅.𝑅
Jobuwanvu bwaK = 𝜆𝑥𝜆𝑑.length(𝑥) ⪰ 𝑑
Jobuwanvu bw’emmeezzaK = 𝜆𝑑.length(𝑡) ⪰ 𝑑
Now let’s return to the exceed comparative we want to analyze, namely (30). Under
the analysis of possessed adjective nominalizations proposed here, the subject and object
of okusinga ‘exceed’ both denote sets of degrees. But these are exactly the arguments that
a 2-place comparative operator expects (cf. 13). If we submit that okusinga has a 2-place
variant as in (39), the analysis of (30) can proceed straightforwardly as in (40)-(41).
(39) J(oku)singa2 K = J-er2 K = 𝜆𝐷1 ∈ 𝐷⟨𝑑,𝑡⟩ 𝜆𝐷2 ∈ 𝐷⟨𝑑,𝑡⟩ .max(𝐷2) ≻ max(𝐷1)
371
M. Ryan Bochnak
(40)
S
DP2
Obuwanvu bw’emmeeza
(41)
a.
b.
c.
d.
e.
VP
V
DP1
bu-singa
obugazi bwayo
JVK = J(oku)singa2 K = J-er2 K = 𝜆𝐷1 ∈ 𝐷⟨𝑑,𝑡⟩ 𝜆𝐷2 ∈ 𝐷⟨𝑑,𝑡⟩ .max(𝐷2) ≻ max(𝐷1)
JDP1 K = 𝜆𝑑.width(𝑡) ⪰ 𝑑
JVPK = 𝜆𝐷2.max(𝐷2) ≻ max(𝜆𝑑.width(𝑡) ⪰ 𝑑)
JDP2 K = 𝜆𝑑.length(𝑡) ⪰ 𝑑
JSK = 1 iff max(𝜆𝑑.length(𝑡) ⪰ 𝑑) ≻ max(𝜆𝑑.width(𝑡) ⪰ 𝑑)
= “the maximal degree to which the table is long is greater than the maximal
degree to which it is wide”
I refer to the second style of analysis for adjective nominalizations as the measure
function analysis. The idea is that instead of denoting relations, nominalized gradable
adjectives instead denote measure functions directly, i.e., functions from individuals to
the maximal degree to which they hold a property (Kennedy 2007b). Under this analysis,
the denotation of obuwanvu ‘length’ would be modeled as in (42).
(42) JobuwanvuK = 𝜆𝑥.length(𝑥)
type ⟨𝑒, 𝑑⟩
Such an analysis has the following consequences. First, in the genitive construction, a
possessed nominalized adjective denotes a degree (instead of a set of degrees), as in (43).
(43) Jobuwanvu bw’emmeezzaK = length(𝑡)
Second, we require a modified lexical entry for 2-place exceed, reflecting this type
difference. Under this analysis, okusinga compares two degrees directly, as in (44). The
derivation in (40) proceeds in the same was as before; it’s only the type of the arguments
of the comparative operator that are different. Also note that there are no maximality
operators within the semantics of oksuinga under this analysis: maximality comes for
free from the measure function itself.
(44) J-singa2 K = 𝜆𝑑 ∈ 𝐷𝑑 𝜆𝑑 ′ ∈ 𝐷𝑑 .𝑑 ′ ≻ 𝑑
372
22 Two-place exceed comparatives in Luganda
Third, if we continue to assume a relational (type ⟨𝑑, ⟨𝑒, 𝑡⟩⟩) analysis for gradable adjectives, then the nominalizing bu- morphology now has the function of turning the
gradable adjective into the corresponding measure function, as in (45).8
(45)
Jbu-K = 𝜆𝐺 ∈ 𝐷⟨𝑑,⟨𝑒,𝑡⟩⟩ 𝜆𝑥.m𝐺 (𝑥)
where m𝐺 is the measure function associated with a gradable predicate 𝐺
Finally, the measure function analysis allows for a more straightforward analysis of
nominalized adjectives in argument position, an example of which is given in (46).
(46)
Obuwanvu bwa
Lydia bwe
wunyisa.
o-bu-wanvu bu-a
Lydia bu-e
wunyisa
aug-nc14-tall nc14-gen Lydia nc14-cleft surprise
‘Lydia’s height is surprising.’
Intuitively, it is Lydia’s maximal degree of height that is surprising, not any smaller
degree of height. Once again, the measure function analysis of nominalized adjectives
gets maximality for free, whereas the relational analysis must posit an ad hoc maximality
operator to turn the ⟨𝑑, 𝑡⟩ expression into a degree.9
Summarizing, I take evidence from subcomparatives to indicate that Luganda has a
two-place version of the comparative operator, alongside a three-place variant. There
are two plausible analyses of nominalized gradable adjectives that allow for a straightforward analysis of two-place exceed comparatives like (30). While I do not come down
definitively in favor of either the relational or measure function analysis for nominalized
gradable adjectives, examples like (46) may point towards the measure function analysis.
5 Conclusion
I have argued for the existence of (at least) two versions of (oku)singa qua comparative
operator in Luganda, namely those in (47)-(48).10
(47)
8
3-place comparative:
J(oku)singa)3 K = 𝜆𝑥𝜆𝐺𝜆𝑦.max(𝜆𝑑.𝐺(𝑑)(𝑦)) ≻ max (𝜆𝑑 ′ .𝐺(𝑑 ′ )(𝑥))
Alternatively, if we take as a starting point a measure function analysis of gradable adjectives (Bartsch &
Vennemann 1972; Kennedy 2007b), then we maintain that bu- is semantically vacuous:
(i) J-wanvu𝐴 K = Jobuwanvu𝑁 K = 𝜆𝑥.length(𝑥)
9
Of course, this also involves making the non-trivial assumption that predicates like wunyisa ‘surprise’ can
be predicates of degrees. See Castroviejo & Schwager (2008) and Moltmann (2009) for discussion of related
issues.
10 Which version of the 2-place operator in (48) we choose depends on whether we adopt the relational or
measure function analysis of nominalized gradable adjectives.
373
M. Ryan Bochnak
(48)
2-place comparative:
J(oku)singa2 K = 𝜆𝐷1𝜆𝐷2.max(𝐷2) ≻ max(𝐷1)
or
J(oku)singa2 K = 𝜆𝑑𝜆𝑑 ′ .𝑑 ′ ≻ 𝑑
Significantly, Luganda has a 2-place exceed comparative despite having only syntactically “phrasal” standards. This results in an important consequence for theories of comparatives cross-linguistically, namely that having only syntactically phrasal standards
does not necessarily entail the absence of 2-place comparatives. Meanwhile, the right
analysis for nominalization depends on our starting assumptions about the underlying
meaning of gradable adjectives and the semantic type of possessed adjective nominalizations.
Acknowledgments
I thank Kisuule Magala Katende for his patience and enthusiasm for working with me on
his language. I also thank Karlos Arregi, Chris Kennedy, Jason Merchant, two anonymous
reviewers, as well as audiences at the University of Chicago, UC Berkeley, ACAL 45 at the
University of Kansas, and the Semantics of African, Asian, and Austronesian Languages
at the University of Tübingen for comments and criticisms. The usual disclaimers apply.
Abbreviations
aug
cleft
gen
loc
augment
cleft
genitive marker
locative
nc#
poss
pst
noun class
possessive
past
References
Barker, Chris. 1995. Possessive descriptions. Stanford, CA: CSLI Publications.
Bartsch, Renate & Theo Vennemann. 1972. The grammar of relative adjectives and comparison. Linguistische Berichte 20. 19–32.
Beck, Sigrid, Sveta Krasikova, Daniel Fleischer, Remus Gergel, Stefan Hofstetter, Christiane Savelsberg, John Vanderelst & Elisabeth Villalta. 2009. Crosslinguistic variation in comparative constructions. In Jeroen van Craenenbroeck & Johan Rooryck
(eds.), Linguistic Variation Yearbook, vol. 9, 1–66. Philadelphia: John Benjamins.
DOI:10.1075/livy.9.01bec
Beermann, Dorothee, Jonathan Brindle, Lars Hellan, Solomon Telda, Janicke Furberg,
Florence Bayiga & Yvonne Otoo. 2005. A comparison of comparatives. In Miriam Butt
& Tracy Holloway King (eds.), Proceedings of LFG05, 42–53. Stanford, CA: CSLI Publications.
374
22 Two-place exceed comparatives in Luganda
Bhatt, Rajesh & Shoichi Takahashi. 2011. Reduced and unreduced phrasal comparatives.
Natural Language and Linguistic Theory 29. 581–620.
Bochnak, M. Ryan. 2013. Cross-linguistic variation in the semantics of comparatives.
Chicago: University of Chicago dissertation.
Castroviejo, Elena & Magdalena Schwager. 2008. Amazing DPs. In Tova Friedman &
Satoshi Ito (eds.), Proceedings of Semantics and Linguistic Theory (SALT) XVIII, 176–193.
Ithaca, NY: CLC Publications.
Chomsky, Noam. 1977. On wh-movement. In Peter Culicover, Thomas Wasow & Adrian
Akmajian (eds.), Formal syntax, 71–132. New York: Academic Press.
Cresswell, Max J. 1976. The semantics of degree. In Barbara Partee (ed.), Montague Grammar, 261–292. New York: Academic Press.
Heim, Irene. 1985. Notes on comparatives and related matters. Ms. University of Texas
at Austin.
Heim, Irene. 2001. Degree operators and scope. In Caroline Féry & Wolfgang Sternefeld
(eds.), Audiatur Vox Sapientiae, 214–239. Berlin: Akademie-Verlag.
Howell, Anna. 2013. Abstracting over degrees in Yoruba comparison constructions. In
Emmanuel Chemla, Vincent Homer & Grégoire Winterstein (eds.), Proceedings of Sinn
und Bedeutung 17, 271–288. Paris. Talk presented at Sinn und Bedeutung 17.
Kennedy, Christopher. 1997. Projecting the adjective: The syntax and semantics of gradability and comparison. University of California, Santa Cruz dissertation.
Kennedy, Christopher. 2007a. Modes of comparison. In Malcolm Elliott, James Kirby, Osamu Sawada, Eleni Staraki & Suwon Yoon (eds.), Chicago Linguistic Society (CLS) 43.
Chicago: Chicago Linguistic Society.
Kennedy, Christopher. 2007b. Vagueness and grammar: The semantics of relative and absolute gradable adjectives. Linguistics and Philosophy 30(1). 1–45. DOI:10.1007/s10988006-9008-0
Kennedy, Christopher & Louise McNally. 2005. Scale structure, degree modification and the semantics of gradable predicates. Language 81(2). 345–381.
DOI:10.1353/lan.2005.0071
Lechner, Winfried. 2001. Reduced and phrasal comparatives. Natural Language and Linguistic Theory 19(4). 683–735.
Lin, Jo-wang. 2009. Chinese comparatives and their implicational parameters. Natural
Language Semantics 17. 1–27.
Merchant, Jason. 2009. Phrasal and clausal comparatives in Greek and the abstractness
of syntax. Journal of Greek Linguistics 9. 134–164.
Moltmann, Friederike. 2009. Degree structure as trope structure: A trope-based analysis
of positive and comparative adjectives. Linguistics and Philosophy 32. 51–94.
Nicolas, David. 2004. The semantics of nouns derived from gradable adjectives. In Cécile
Meier & Matthias Weisgerber (eds.), Proceedings of Sinn und Bedeutung 8, 197–207.
von Stechow, Arnim. 1984. Comparing semantic theories of comparison. Journal of Semantics 3. 1–77. DOI:10.1093/jos/3.1-2.1
375
Chapter 23
Temporal remoteness and vagueness in
past time reference in Luganda
M. Ryan Bochnak
University Konstanz
Peter Klecha
The Ohio State University
In this paper, we point out that past time operators (PTOs) in Luganda, a language that
makes three past time remoteness distinctions, are vague and context-dependent, and provide an analysis whereby PTOs contain context-sensitive measure functions akin to gradable
adjectives. We call the relevant PTOs RECENT, INTERMEDIATE, and DISTANT, respectively. Luganda PTOs give rise to borderline cases, where it is difficult to decide whether a
past reference time (RT) counts as ‘recent’, ‘intermediate’ or ‘distant’. What counts as ‘recent’, ‘intermediate’ or ‘distant’ is context dependent; e.g., there are contexts where REC is
acceptable with an RT of a few weeks ago, and contexts where DIST is acceptable for an
RT of a few minutes ago. We assume that like tenses in English, PTOs in matrix clauses in
Luganda restrict the relation between utterance time (UT) and RT. However, while English
past tense presupposes that RT precedes UT (e.g. Kratzer 1998), Luganda PTOs additionally
encode as part of their meaning a vague, context-dependent measure function that compares
the length of a time interval to a contextual standard.
1 Introduction
So-called ‘graded tense’ systems are cross-linguistically quite common (Comrie 1985;
Dahl 1983; Mithun 1999; Nurse 2008). Such systems make use of multiple morphemes that
make more fine-grained distinctions than simply ‘past’ or ‘future’. Rather, they encode
varying degrees of remoteness (e.g., recent vs. distant) from a reference point. Languages
with such systems vary with respect to what distinctions are made, and how many.
However, grammaticalized temporal remoteness morphemes have only recently begun to attract attention within formal semantics (see e.g., Cable 2013 for Gĩkũyũ, Hayashi
2011 for South Baffin Inuktitut; Mucha 2014 for Medumba). We aim to expand upon this
discussion here by investigating the paradigm of graded past tenses in Luganda (Bantu;
M. Ryan Bochnak & Peter Klecha. Temporal remoteness and vagueness in past time reference
in Luganda. In Jason Kandybowicz, Travis Major, Harold Torrence & Philip T. Duncan (eds.),
African linguistics on the prairie: Selected papers from the 45th Annual Conference on African
Linguistics, 377–391. Berlin: Language Science Press. DOI:10.5281/zenodo.1251752
M. Ryan Bochnak & Peter Klecha
JE15). Luganda has three remoteness categories for past time reference. We refer to these
as recent (1), intermediate (2), and distant (3). They are bound morphemes that are
obligatory on verbs with past time reference, i.e., a speaker must choose one of these
three forms for finite clauses with past time reference.
(1)
(2)
(3)
Nzinye
(ku matya).
1sg.dance.rec.pst (loc morning)
‘I danced (this morning).’
Template (rec.pst): agr-∅-root-ie
Nazinye
(jjo).
1sg.dance.int.pst (yesterday)
‘I danced (yesterday).’
Template (int.pst): agr-a-root-ie
Nazina
(luli).
1sg.dance.dist.pst (another.time)
‘I danced (the other day).’
Template (dist.pst): agr-a-root-a
recent past
intermediate past
distant past
(modulo irregular verbs)
The boundaries of the temporal delineations are often described in the literature as
being quite precise, e.g., up to 4 hours ago; one day ago. In this connection, it is quite
common for authors to use labels such as ‘hesternal’ or ‘hodiernal’. Despite such characterizations, it is also often reported that the use of these morphemes can be rather flexible
and context-sensitive (e.g., Hyman 1980 on Bamileke; Sharman 1956 on ChiBemba). Our
primary objective in this paper is to show that the temporal remoteness morphemes1
(TRMs) in Luganda display the hallmarks of vagueness found elsewhere in unrelated phenomena across languages (e.g., certain nouns and gradable adjectives); in other words,
in certain contexts, even native speakers with full knowledge of the facts are unable to
determine which TRM should be used. We provide an analysis of TRMs in Luganda that
takes into account the properties of vagueness we observe. In doing so, we also compare
the behavior of temporal operators cross-linguistically with that of relative and absolute
gradable adjectives.
The paper proceeds as follows. In §2 we describe in more detail the use of TRMs in Luganda. In §3 we outline the behavior of vague expressions in general, and in §4 we show
that Luganda TRMs indeed display the properties of vague expressions. Our analysis is
presented in §5, and in §6 we draw a comparison between types of temporal operators
on one hand and gradable adjectives on the other. §7 concludes and hints at areas for
further research.
1 Following
Cable (2013), we use the term ‘temporal remoteness morphemes’ so as not to prejudge their
analysis as true tenses or something else, e.g., modifiers of events as Cable argues for Gĩkũyũ. We do,
however, analyze them as tenses in §5.
378
23 Temporal remoteness and vagueness in past time reference in Luganda
2 Temporal remoteness morphemes in Luganda
Recall that Luganda has three remoteness categories in the past tense paradigm, which
we label as recent, intermediate, and distant past in (1)-(3).2 Ashton et al. (1954: 122–
123) characterize these morphemes in the following way: the recent past “expresses the
completion of an action and/or state entered upon within the immediate past”; the intermediate past “expresses an action actually finished and accomplished as an action, but
confined or limited … , roughly speaking, to the past twelve hours”; and the distant past
“denotes an action in the past, but is indefinite as to the exact time. … It corresponds to
a Past Aorist.” We spend time in this section going into more detail about the semantics
and use of these morphemes.3
First, we argue that these morphemes are not aspectual, since they can freely combine with predicates of all Aktionsarten without coercing any aspectual interpretation.
Witness (4)-(7).
(4) Statives:
a. Babilie abadde
lubuto.
B.
3sg.be.rec.pst nc11.pregnant.woman
‘Babilie was just pregnant.’
b. Babilie yabadde
lubuto.
B.
3sg.be.int.pst nc11.pregnant.woman
‘Babilie was pregnant (a little while ago).’
lubuto.
c. Babilie yali
3sg.be.dis.pst nc11.pregnant.woman
B.
‘Babilie was pregnant (a long time ago).’
(5) Activities:
a. Kato azinye.
K. 3sg.dance.rec.pst
‘Kato just danced.’
b. Kato yazinye.
K. 3sg.dance.int.pst
‘Kato danced (a little while ago).’
2
Our data come from elicitation sessions with two native speakers of Luganda, bilingual in English, living
in the United States. All elicitation sessions were conducted in English, and involved a mix of translation
tasks and felicity judgements of grammatical sentences in particular contexts.
3 Note that we restrict ourselves here to the three “simple” past time markers in matrix clauses. We also do not
talk about so-called “special” and “compound” tenses of Luganda (see Kisubika-Musoke 1986). Grammars
such as Ashton et al. (1954) have also cited future TRMs which similarly divide future time into nearer and
more distant time zones. However, our consultants have indicated to us that these forms are not colloquial,
and the commonly used future forms do not carry temporal remoteness inferences, so we do not discuss
these here.
379
M. Ryan Bochnak & Peter Klecha
c. Kato yazina.
K. 3sg.dance.dis.pst
‘Kato danced (a long time ago).’
(6) Accomplishments:
a. Kato addusse
mairo biri.
K. 3sg.run.rec.pst mile three
‘Kato just ran three miles.’
b. Kato yaddusse
mairo biri.
K. 3sg.run.int.pst mile three
‘Kato ran three miles (a little while ago).’
c. Kato yadduka
mairo biri.
K. 3sg.run.dis.pst mile three
‘Kato ran three miles (a long time ago).’
(7)
Achievements:
a. Ensuwa
eyatisse.
aug.nc9.pot nc9.break.rec.pst
‘The pot just broke.’
b. Ensuwa
yeyatisse.
aug.nc9.pot nc9.break.int.pst
‘The pot broke (a little while ago).’
c. Ensuwa
yeyatika.
aug.nc9.pot nc9.break.dist.pst
‘The pot broke (a long time ago).’
Second, certain TRMs in Luganda asymmetrically entail others. In particular, the intermediate past appears to be usable in any context where the recent past is. For instance
in (8)-(9), the intermediate past is possible both with tano emabega ‘five hours ago’ and
satu emabega ‘three hours ago’, while the recent past is possible only with the latter.
(8)
(9)
Nzinye
saawa {satu/#tano} emabega.
1sg.dance.rec.pst hour {three/five} behind
‘I danced {three/five} hours ago.’
Nazinye
saawa {satu/tano} emabega.
1sg.dance.int.pst hour {three/five} behind
‘I danced {three/five} hours ago.’
(recent past)
(intermediate past)
The distant past cannot as easily be used in contexts where the intermediate or recent
past can, but there is some evidence that it is the most general past, and that the inference
of ‘distance’ is due to scalar implicature. For instance, a question with a the recent or
380
23 Temporal remoteness and vagueness in past time reference in Luganda
intermediate past gives rise to an inference that the speaker has at least some knowledge
about the temporal location of the event at issue – but a question with the distant past
gives rise to no such inference. The contexts below are adapted from Cable (2013), who
finds the same thing for Gĩkũyũ.
(10) Context: You have known Kato for a long time, but have never been to his house.
You are finally invited over and you see that he often buys very old things for
fun. He tells you he bought an old Apple computer from 1985 just a few hours
before you arrived. You see his TV, which is also quite old, but you can’t tell if he
just recently bought it, or if he’s actually had it since the 80’s.
a. Eno
TV wagigguladi?
this.nc9 TV 2sg.subj-nc9.obj-buy-dis.pst-when
‘When did you buy this TV?’
b. Naagigguzze
jjo.
1sg.subj.nc9.obj.buy.int.pst yesterday
‘I just bought it yesterday.’
(11) Same context as above.
TV wagigguzzedi?
a. Eno
this.nc9 TV 2sg.subj.nc9.obj.buy.int.pst.when
‘When did you buy this TV?’ (Suggests it happened recently)
b. Naagiggula
luli.
1sg.subj.nc9.obj.buy.dis.pst another.time
‘I bought it a while ago.’ (Sounds contradictory)
c. Speaker notes that one might precede the answer in 11b by saying “What
makes you think I bought it recently?”.
Crucially, the speaker notes that (11a) can be challenged, by saying something like
“What makes you so sure I just bought it?”, whereas (10a) cannot be challenged similarly,
i.e., with something like “What makes you so sure I bought it a long time ago?”. Thus we
conclude that the inference of temporal distance that is produced by what we call the distant past is an implicature. But note that this implicature is vague and context-sensitive
in exactly the same way that the semantic inferences of the recent and intermediate
past are vague and context-sensitive. Thus for the following discussion we will not distinguish between implicatures and semantic inferences when discussing vagueness and
context-sensitivity.
3 Vagueness in natural language
Certain expressions in natural language display vagueness, for instance certain ‘scalar’
nouns like heap, and relative gradable adjectives like tall. Vague expressions are those
where the criteria of application are not clear-cut and can shift in different contexts of
381
M. Ryan Bochnak & Peter Klecha
use. Here we outline three properties of vague expressions and sentences that contain
them (Fine 1975; Kamp 1975; Fara 2000; Kennedy 2007).
First, context dependence: what counts as a heap or tall varies from context to context. Consider the sentence in (12).
(12)
Tom is tall.
In a context where Tom has a height of 165 cm, and is in the second grade, (12) is intuitively true. Meanwhile, in a context where Tom has a height of 165 cm and is a professional basketball player, (12) is intuitively false. Crucial here is the comparison class,
namely the set of objects against which Tom’s height is compared in order to judge
whether (12) counts as true or not.
Second is the existence of borderline cases. That is, there are certain cases where it’s
hard to judge whether a vague predicate holds or not. For instance, consider a context
where Tom is a 12-year-old boy. Is (12) true or false? This case seems harder to judge
one way or the other, compared with the previous contexts we considered where the
judgements were more clear-cut.
The third property of vague expressions is that they give rise to the sorites paradox.
Consider (13).
(13)
P1: 1,000,000 grains of sand is a heap of sand.
P2: A heap of sand less one grain is still a heap of sand.
C: #One grain of sand counts as a heap of sand.
If we accept P1 & P2, then with enough iterations of P2 we should also accept C. Nevertheless, the intuition is that C does not hold, resulting in a paradox.
There are various proposals on the market for how vagueness comes about, and we
do not wish to take a stand on this issue here.4 However, we would like to highlight
the following passage from Kennedy (2007: p. 42): “[V]agueness comes from epistemic
uncertainty about where we actually draw the line and metalinguistic resistance to treating highly similar objects differently relative to the property expressed …Whether this
analysis extends to an account of vagueness in other categories is an issue that must
be addressed in future work.” It is this last point that we aim to address in this paper,
namely whether these properties of vagueness manifest themselves in temporal operators in natural language.
4 Vagueness in the Luganda past paradigm
In this section we show that Luganda TRMs display the same properties of vagueness
that we find in vague nouns and adjectives. First, a methodological point: we presented
discourse contexts to the consultant in English, which does not have TRMs, because we
4
See Fara 2000, Grinsell 2012, Soames 1999 and Williamson 1994 for various proposals.
382
23 Temporal remoteness and vagueness in past time reference in Luganda
wanted to avoid prejudicing the response given by the consultants by having a TRM in
the prompt.5
Context-dependence: for many events, the restrictions on when certain TRMs can be
used more or less follows the pattern described in Ashton et al. (1954). However, what
counts as recent or distant can vary across contexts. For instance, the recent past can be
used to talk about a time months prior to the utterance time (UT), as in (14). Likewise,
the distant past can be used to talk about a time only a few minutes prior to UT, as in
(15).6 Thus, the notions of what counts as recent or distant can vary with the context,
much like what counts as tall varies across contexts.
(14)
a. Context: You plant your crops every year in February. It is now April, and I
ask you what you planted this year. You tell me that you planted maize.
b. Nsimbye
kasooli.
1sg.plant.rec.pst maize
‘I planted maize.’
(15)
a. Context: We are at a party, and I ask you why you are not dancing to the
song that’s playing. You tell me that you danced a few songs ago.
luli.
b. Nazina
1sg.dance.dist.pst another.time
‘I danced a while ago (to another song).’
In other words, the use of a particular TRM seems to depend on a relevant comparison
class, just like for predicates like tall. Here is a first stab at characterizing the relevant
comparison class for TRMs: in (14) the comparison class is time intervals between UT
and previous crop plantings, while in (15) the comparison class is time intervals between
UT and previous songs.
Borderline cases: there are also cases where it is difficult to decide whether an event
counts as ‘recent’, ‘intermediate’ or ‘distant’. There is an interesting contrast here between TRMs in Luganda and gradable adjectives. Namely, for gradable adjectives, a
speaker can refuse to make a judgment (e.g. “I’m not sure whether Bill would count
as tall or not”). However, verbs in Luganda naming past events are obligatorily marked
with TRMs - a speaker always has to choose one. Consider the following responses to
the question “What have you been doing?”
(16) Nzinye
saawa {satu/?nnya/#tano} emabega.
1sg.dance.rec.pst hour {three/four/five} behind
‘I danced {three/four/five} hours ago.’
(17) Nazinye saawa {satu/nnya/tano} emabega.
‘I danced {three/four/five} hours ago.’
5
6
(recent past)
(intermediate past)
See AnderBois & Henderson (2015) for comments on choosing elicitation language in semantic fieldwork.
One of our consultants finds (15) not completely natural, but does accept it.
383
M. Ryan Bochnak & Peter Klecha
(18)
#Nazina saawa {satu/nnya/tano} emabega.
‘I danced {three/four/five} hours ago.’
(distant past)
According to our consultants, in the context in (16), the use of the recent past is definitely fine for ‘3 hours ago’, definitely excluded for ‘5 hours ago’, but borderline for ‘4
hours ago’.7
Sorites sequences: Luganda TRMs also give rise to sorites paradox effects, just like
scalar nouns and gradable adjectives. Consider the party scenario in (19), and the subsequent sentences offered and rejected by our consultant to describe the scenario in
(20)-(23).8
(19) Context: You are at a party, and they play the following songs (in order):
Twist and Shout
YMCA
Dancing Queen
Sweet Caroline
Gangnam Style
Blue Suede Shoes
Don’t Stop Believing
Macarena
Rollin’ in the Deep
Zamboni Driver
(20)
a. Context as in (19). Peet comes to the party late, just as Zamboni Driver is
finishing. He is always really interested in the music they play, so you want
to tell him what songs were played.
b. Baakubye
Twist and Shout.
3pl.play.int.pst Twist and Shout
‘They played Twist and Shout.’
c. Observation: Speaker uses int.pst discourse-initially in this scenario.
(21)
a. Context: same as above
b. # Baakuba
Twist and Shout.
3pl.play.dist.pst Twist and Shout
‘They played Twist and Shout.’
c. Speaker’s comment: “Sounds like it’s a different party, another day.”
(22)
a. Context: same as above
b. # Bakubye
Twist and Shout.
3pl.play.rec.pst Twist and Shout
‘They played Twist and Shout.’
7 Given
the variable nature of context dependency and borderline cases, we imagine that judgments could
also vary somewhat between speakers.
8 We have performed this activity with only one of our consultants so far (Kisuule Magala Katende).
384
23 Temporal remoteness and vagueness in past time reference in Luganda
c. Speaker’s comment: “That’s like they just played it. That’s like a couple of
minutes ago. … Just missed it.”
(23)
a. Context: Immediately after uttering (20b)
b. Baaziza-ko
YMCA.
3pl.add.int.pst-part YMCA
‘Then they played YMCA.’
(‘They added on YMCA.’)
(offered by speaker)
c. Observation: Speaker uses int.pst again for event immediately following the
first
(24)
a. Context: Immediately after uttering (20b) and (23b)
b. Baazizako Dancing Queen.
‘Then they played Dancing Queen.’
int.pst
Judgment: Accepted by speaker
c. Baazizako Sweet Caroline.
‘Then they played Sweet Caroline.
int.pst
Judgment: Accepted by speaker
⋮
d. Baazizako Zamboni Driver.
‘Then they played Zamboni Driver.’
int.pst
Judgment: Accepted by speaker
e. Speaker’s comment: “If you just played it, you would say bazaako [rec.pst].
… When you are coming towards the end, it has to be bazaako. … Because
it’s more recent than the other ones.”
f. Peet: “Can you be really sure about where you would switch?”
Speaker: “No, the time difference is in your mind. … There is no obvious
timeline to stop it.”
A similar effect is observed in the following OCD friend scenario in (25)-(27).
(25)
a. Context: Your (crazy) friend wants to know what songs were played on the
radio in the last few days in order. Being an indulgent friend, you tell him
what songs were played.
b. Baakuba
Twist and Shout.
3.sg.play.dist.pst Twist and Shout
‘They played Twist and Shout.’
c. Speaker’s comment: “That’s another day. That’s not today.”
(26)
a. Context: Immediately after uttering (25b)
b. Baakuba
YMCA.
3.sg.play.dist.pst YMCA
‘They played YMCA.’
385
M. Ryan Bochnak & Peter Klecha
(27)
a. Context: Immediately after uttering (25b)
b. # Baakubye
YMCA.
3.sg.play.int.pst
‘They played YMCA.’
c. Speaker’s comment: “No. It would be very awkward. … You never know
when you switch to baakubye [int.pst]. But that should be within a day or
something like that. … There’s no clear place where you would stop to apply
it [the distant past form].”
Thus, we seem to have evidence that sorites sequences behave like expected for sentences containing vague expressions. Small differences in time don’t prompt speakers
to suddenly switch TRMs, and speakers are not sure when exactly it is appropriate to
switch TRMs in these listing contexts.9
In sum, sentences containing TRMs in Luganda behave like sentences containing other
vague expressions (in English).
5 Analysis
Following Kennedy (1997; 2007), we take simple relative gradable adjectives like tall,
which in their positive (i.e., bare) forms are vague, to denote measure functions, i.e.,
functions from an object to a degree, an abstract unit of measurement. The denotation of
tall on this basis is given below.
(28)
J tall K = 𝜆𝑥.height(𝑥)
type ⟨𝑒, 𝑑⟩
The above says that tall maps an individual to her height. Any simple adjective phrase,
however, should ultimately denote something of type ⟨𝑒, 𝑡⟩; i.e., should map an individual
to true or false, depending on her height. So some assumption must be made about how
the type of a bare adjective shifts when in its positive form. Kennedy’s proposal is an
abstract morpheme pos which has the following interpretation.
(29)
J pos K = 𝜆𝐺⟨𝑒,𝑑⟩ .𝜆𝑥.𝐺(𝑥) ⪰ s(𝐺)
type ⟨⟨𝑒, 𝑑⟩, ⟨𝑒, 𝑡⟩⟩
Following Kennedy (2007), the function s is a contextually provided one which takes
a measure function and returns the degree that that function would have to map an
individual to in order for that individual to ‘stand out’, in terms of that property, in the
context. If tall combines with pos rather than a degree modifier, the result is a vague
predicate.
(30) J pos tall K = 𝜆𝑥.height(𝑥) ≻ 𝑠
9 It
type ⟨𝑒, 𝑡⟩
has been suggested to us that these effects might be obviated if the list is given out of order. We have
not been able to test this idea yet.
386
23 Temporal remoteness and vagueness in past time reference in Luganda
Thus a sentence like John is tall, which includes the positive form of an adjective, is
true iff John’s height stands out in the context, i.e., if it exceeds the determined standard.
On Kennedy’s account, borderline cases arise because speakers are unsure about where
the cut-off is for ‘standing out’. Cases like the Sorites Paradox arise because given a context that contains any two objects that have gradable property 𝐺 to very slightly varying
degrees, neither one can ‘stand out’ with respect to the other.
Note that the reason that tall is not simply given the denotation in (30) above is that
it may instead combine with degree modifiers or measure phrases, which do not necessarily involve comparison of superiority. However, Luganda TRMs do not compose
with anything like degree modifiers, and so we need not commit to any account of the
positive form of gradable adjectives.10 We provide denotations for the Luganda TRMs
below which build in the vagueness witnessed in the positive morpheme above.
Our denotations are built upon the following measure functions, which relate time
intervals to degrees.
(31) close(𝑡, 𝑡 ′ ) assigns to a time 𝑡 ′ a degree on a scale of closeness to 𝑡
(32) far(𝑡, 𝑡 ′ ) assigns to a time 𝑡 ′ a degree on a scale of distance from 𝑡
We embed our analysis in a theory of tense which says that, like tenses in English,
TRMs in Luganda in matrix clauses constrain the relation between an evaluation time
and a reference time (RT), where that evaluation time is, at least in matrix contexts,
utterance time (Reichenbach 1947; Klein 1994). However, cf. Cable (2013; 2015) for a different view of TRMs in Gĩkũyũ. Following Heim (1994) and Kratzer (1998) we assume
that temporal operators constrain the relation between RT and UT by placing presuppositions on the reference of a temporal pronoun which corresponds to RT. Following
Cable (2013) and others, we assume that TRMs do not denote this pronoun itself, but
adjoin to such pronouns. Note that these are all assumptions, and moreover our analysis
does not crucially hinge on any of them.
(33)
J rec.pst K𝑡 = 𝜆𝑡 ′ ∶ close(𝑡, 𝑡 ′ ) ≻ s(close) & 𝑡 ′ < 𝑡 . 𝑡 ′
Presupposition: the degree of closeness of 𝑡 ′ to 𝑡 exceeds a contextual standard
(34) J int.pst K𝑡 = 𝜆𝑡 ′ ∶ far(𝑡, 𝑡 ′ ) ≺ s(far) & 𝑡 ′ < 𝑡 . 𝑡 ′
Presupposition: the degree of far-ness of 𝑡 ′ to 𝑡 is less than a contextual standard
Implicature: the degree of closeness of 𝑡 ′ to 𝑡 is less than a contextual standard
(35) J dist.pst K𝑡 = 𝜆𝑡 ′ ∶ 𝑡 ′ < 𝑡 . 𝑡 ′
Implicature: the degree of far-ness of 𝑡 ′ to 𝑡 exceeds a contextual standard.11
10
A reviewer suggests that temporal adverbs might be treated as degree modifiers under this view. We leave
this intriguing suggestion for future work.
11
Crucially the scalar implicature generated by comparison to the other tenses carries the same vagueness
and context-dependence that those tenses do.
387
M. Ryan Bochnak & Peter Klecha
In the denotations above, 𝑡 ′ corresponds to RT, i.e., the temporal pronoun that the
TRMs adjoin to, while 𝑡 corresponds to the evaluation time, which in matrix contexts is
UT.12
Each of these morphemes is a partial identity function on RT, presupposing two things
about it: First, that it is in the past relative to UT (𝑡 ′ < 𝑡), and second, its distance from
𝑡, denoted by the measure function and the comparative operator (≻ or ≺).
Our analysis of these TRMs makes them comparable to the positive forms of relative
gradable adjectives like tall. So our analysis captures the vagueness associated with these
expressions. Note that this analysis does not commit us to one view of vagueness over
another; since our analysis equates these expressions with positive gradable adjectives,
whatever account can be made of gradable adjectives can be extended to TRMs.
6 Accounting for crosslinguistic variation: scale type and
vagueness
Why are Luganda TRMs vague while English tenses are not? We suggest it is for the same
reason that relative gradable adjectives are vague in their positive forms, while absolute
gradable adjectives are not. Absolute adjectives, like full, are still perfectly gradable (36),
but fail the tests described above for vagueness.
(36)
This cup is more full than that one.
For example, for (37) to be true, the degree of fullness exhibited by the cup must simply
be the maximum; there is no contextually determined standard.
(37)
This cup is full.
Likewise, such adjectives do not have borderline cases or give rise to Sorites Paradoxes.13 Kennedy (2007) attributes this to the differences between these adjectives in
terms of their scale structure. In other words, while there is an inherent upper-bound to
fullness, there is no inherent upper-bound for height. So for an object to ‘stand-out’ on
the fullness scale it simply must occupy the highest point on said scale, while the same
cannot be said for the height scale. Instead, since such scales lack obvious ‘milestones’
for determining what stands out, interlocutors must appeal to context. This in turn gives
rise to vagueness.
English past tense is like an absolute gradable adjective; it does not exhibit vagueness
because it does not depend upon context (not in the same way as a relative gradable
adjective, anyway). We therefore predict that no language with a single past tense like
English should exhibit vagueness either. We also predict that, if any languages with
graded tense systems like Luganda actually do differentiate them in a rigid, diurnal way,
12
The same may also be true for embedded contexts; see Cable (2013)
that, as Kennedy (2007) discusses, such adjectives may give rise to imprecision, e.g., in a case where a
cup which is not strictly speaking full is called ‘full’ in a particular context; this is different from vagueness.
See Kennedy 2007 for more discussion.
13 Note
388
23 Temporal remoteness and vagueness in past time reference in Luganda
as Luganda was described to do by Ashton et al. (1954), then we also predict that those
graded tense systems should not exhibit vagueness either. And finally, we predict that
any graded tense system which is context dependent like Luganda’s should also be vague.
7 Conclusion and future research
Contrary to descriptions like that of Ashton et al. (1954), the graded tense system of
Luganda exhibits context dependence and vagueness. This finding expands the empirical
domain for work on vagueness, which so far has been focused mostly on adjectives and
other lexical categories.
Many questions remain. For example, Cable (2013) argues against treating TRMs in
Gĩkũyũ as tenses, relying on data from adverbials and embedding. Tenses, according to
Klein (1994), relate reference time to utterance time, while the relation between reference
time and the event described by the verb is mediated by aspect. According to Cable, TRMs
in Gĩkũyũ behave like neither tense nor aspect according to these definitions, instead
relating utterance time directly to the event described by the verb. It remains to be seen
whether Luganda TRMs are like those in Gĩkũyũ in this regard. This question is left for
future research.
Abbreviations
1, 2, 3
agr
aug
dist
int
loc
nc
first, second, third person
agreement
augment
distant
intermediate
locative
noun class
obj
part
pst
rec
sg
subj
object
particle
past
recent
singular
subject
Acknowledgements
We thank Kisuule Magala Katende and Waiswa Nkwanga for their patience and enthusiasm for working with us on their language. We also thank audiences at the University of Chicago, the University of Potsdam, SWAMP at University of Michigan, the 40th
Berkeley Linguistics Society, ACAL 45 at the University of Kansas, and two anonymous
reviewers for comments and criticisms. The usual disclaimers apply.
389
M. Ryan Bochnak & Peter Klecha
References
AnderBois, Scott & Robert Henderson. 2015. Linguistically establishing discourse context: Two case studies from Mayan languages. In M. Ryan Bochnak & Lisa Matthewson (eds.), Methodologies in semantic fieldwork, 207–232. New York: Oxford University
Press.
Ashton, Ethel O., Enoch M. K. Mulira, E. G. M. Ndawula & Archibald N. Tucker. 1954. A
Luganda Grammar. New York: Longmans, Green & Co.
Cable, Seth. 2013. Beyond the past, present, and future: Towards the semantics of ‘graded
tense’ in Gı̃kũyũ. Natural Language Semantics 21. 219–276.
Cable, Seth. 2015. Graded tenses in complement clauses: Evidence that the future is not
a tense. In NELS 45.
Comrie, Bernard. 1985. Tense. New York: Cambridge University Press.
Dahl, Östen. 1983. Temporal distance: Remoteness distinctions in tense-aspect systems.
Linguistics 21(1). 105–122.
Fara, Delia Graff. 2000. Shifting sands: An interest-relative theory of vagueness.
Philosophical Topics 28. 45–81. Originally published under the name “Delia Graff”.
DOI:10.5840/philtopics20002816
Fine, Kit. 1975. Vagueness, truth and logic. Synthese 30. 265–300.
Grinsell, Timothy. 2012. Social choice theory and linguistic vagueness. Paper presented
at Semantics and Linguistic Theory (SALT) XXII.
Hayashi, Midori. 2011. The structure of multiple tenses in Inuktitut. University of Toronto
dissertation.
Heim, Irene. 1994. Comments on Abusch’s theory of tense. In Hans Kamp (ed.), Ellipsis,
tense, and questions, 143–170. Amsterdam: University of Amsterdam.
Hyman, Larry M. 1980. Relative time reference in the Bamileke tense system. Studies in
African Linguistics 11(2). 227–237.
Kamp, Hans. 1975. Two theories of adjectives. In Edward Keenan (ed.), Formal semantics
of natural language, 123–155. Cambridge: Cambridge University Press.
Kennedy, Christopher. 1997. Projecting the adjective: The syntax and semantics of gradability and comparison. University of California, Santa Cruz dissertation.
Kennedy, Christopher. 2007. Vagueness and grammar: The semantics of relative and absolute gradable adjectives. Linguistics and Philosophy 30(1). 1–45. DOI:10.1007/s10988006-9008-0
Kisubika-Musoke, E. M. 1986. Tense formation in Luganda and some problems related to
learning English. Kampala: Uganda Bookshop.
Klein, Wolfgang. 1994. Time in language. New York: Routledge.
Kratzer, Angelika. 1998. More structural analogies between pronouns and tenses. In Proceedings of Semantics and Linguistic Theory (SALT) 8. Ithaca, NY: CLC Publications.
Mithun, Marianne. 1999. The Languages of Native North America. New York: Cambridge
University Press.
Mucha, Anne. 2014. Past interpretation and “graded tense” in Medumba. Ms. University
of Potsdam.
390
23 Temporal remoteness and vagueness in past time reference in Luganda
Nurse, Derek. 2008. Tense and aspect in Bantu. New York: Oxford University Press.
Reichenbach, Hans. 1947. The tenses of verbs. In Hans Reichenbach (ed.), Elements of
symbolic logic, 287–298. New York: The Macmillan Company.
Sharman, J. C. 1956. The tabulation of tenses in a Bantu language (Bemba: Northern
Rhodesia). Africa 26. 29–46.
Soames, Scott. 1999. Understanding truth. Oxford: Oxford University Press.
Williamson, Timothy. 1994. Vagueness. London: Routledge.
391
Chapter 24
Focus marking in Kuria
Meredith Landman
Pomona College
Rodrigo Ranero
University of Maryland, College Park
This paper examines focus marking in Kuria. We propose an account of the syntax/semantics
of the prefix /ne-/. This prefix displays a varied syntactic distribution, posing a puzzle as to
what semantically unifies all of its uses. In focus constructions, /ne-/ obligatorily appears
on a fronted (i.e. focused) phrase, whereas in simple declaratives, /ne-/ obligatorily appears
pre-verbally. Following previous analyses of similar markers in Bantu (Schwarz 2007 for
Kikuyu and Abels & Muriungi 2008 for Kiitharaka), we analyze /ne-/ uniformly as a focus
marker that arises in a focus phrase in the left periphery. We support this account of /ne/ by presenting novel data that suggest that even when /ne-/ occurs pre-verbally, it still
marks focus (VP focus or sentential focus.) We also show how /ne-/ differs syntactically
from similar markers in other languages. For example, Kuria allows for the focus marker to
appear internal to the focused constituent, in contrast with data from Kiitharaka (Abels &
Muriungi 2008) and other languages (see Hartmann & Zimmermann 2009 for Guruntum).
This paper thus discusses a range of data patterns relating to the Kuria prefix /ne-/, offering
insight into a syntax/semantic puzzle as well as cross-linguistic variation in the realization
of focus.
1 Introduction
Kuria (Narrow Bantu E.43) is an understudied language spoken in Kenya and Tanzania.
In this paper, we investigate the syntax and semantics of the Kuria morpheme /ne-/,
which in the literature has been labeled a focus marker (Cammenga 2004; Mwita 2008).
In declaratives, this morpheme obligatorily occurs pre-verbally, as in (1):1,2
1 The
morpheme /ne-/ displays predictable allomorphy, surfacing as either [ne] or as [n], and it is possible
that it is /n-/ that is in fact underlying.
2 In the declaratives presented in this paper, /ne-/ is obligatory only in the remote past and remote future
tenses; /ne-/ does not appear in the immediate past or immediate future tenses—see also Mwita 2008, who
reports that /ne-/ must appear in 11 out of 22 TAM combinations. It is unclear to us why /ne-/ is absent in
certain tenses; this may be due to phonological, syntactic, or semantic/pragmatic conditioning, and we do
not attempt an account of this pattern here.
Meredith Landman & Rodrigo Ranero. Focus marking in Kuria. In Jason Kandybowicz, Travis
Major, Harold Torrence & Philip T. Duncan (eds.), African linguistics on the prairie: Selected
papers from the 45th Annual Conference on African Linguistics, 393–412. Berlin: Language Science
Press. DOI:10.5281/zenodo.1251754
Meredith Landman & Rodrigo Ranero
(1)
Ichi-ng’iti *(n-)cha-a-it-ir-e
ege-toocho.
10-hyena (foc-)10sa-pst-kill-prf-fv 7-rabbit
‘The hyenas killed the rabbit.’
Previous analyses of Kuria /ne-/, as well as similar morphemes in other Bantu languages, treat these items in different ways: Some view them as focus markers (see Cammenga 2004 and Mwita 2008 for Kuria, Schwarz 2007 for Kikuyu, and Abels & Muriungi
2008 for Kiitharaka), while others see them as assertion or polarity markers (see Bergvall
1987 for Kikuyu and Eslinger 2013 for Kuria). We argue here, based on original data, that
/ne-/ is in fact a focus marker, and we propose an extension of Schwarz’s (2007) syntactic account of Kikuyu to Kuria, by which /ne-/ heads a Focus Phrase projection in the
left periphery of the clause.3 Thus, our aims in this paper are twofold: first, empirically,
to document an array of patterns related to Kuria /ne-/, which illustrate the language’s
strategy for focus realization, and, second, from a theoretical perspective, to show that
an extension of a previous syntactic analysis of a similar morpheme in a related language
(i.e., Schwarz’s account of Kikuyu) can capture the Kuria data.
The remainder of this paper is organized as follows. In §2, we present the basic syntactic distribution of /ne-/. In §3, we present evidence for analyzing /ne-/ semantically as a
focus marker. In §4, we propose a syntactic account of /ne-/, extending Schwarz (2007)’s
account of Kikuyu to Kuria. In §5, we compare focus marking in Kuria with what has
been observed for focus marking in other languages, discussing how Kuria fits into a
crosslinguistic typology of focus marking. In §6, we document a number of additional
patterns regarding /ne-/, articulating several issues regarding /ne-/ for future research.
Finally, §7 concludes our paper.
2 The basic distribution of /ne-/
Kuria displays default SVO word order, with some freedom of object ordering in ditransitive and tritransitive constructions. As shown in (1), in declaratives, /ne-/ obligatorily
appears pre-verbally and only once per clause (parallel to Kikuyu; see Schwarz 2007:
142).4,5
3
Mwita (2008) suggests that some cases involving pre-verbal /ne-/ are due to grammaticalization, and therefore lose the focus meaning. We argue here that because some instances of pre-verbal /ne-/ involve focus,
the simplest synchronic analysis is to take all instances of pre-verbal /ne-/ as instantiating focus.
4
This also holds for embedded declaratives, e.g. clauses embedded by a bridge verb such as ‘say’:
(i)
N-eng’we a-a-gamb-er-e
iga gati *(n-)a-a-ha-y-e
umw-igia ege-tabo.
foc-who sa-pst-say-prf-fv comp 1.Gati foc-3sg.sa-pst-give-appl.prf-fv 1-teacher 5-book
‘Who said that Gati gave the teacher a book?’
5
See, however, §5.1, in which we observe some phrase-internal instances of /ne-/.
394
24 Focus marking in Kuria
In wh-questions and focus constructions, that is, constructions that involve a fronted
constituent, /ne-/ obligatorily precedes the fronted constituent. For example, in the whquestions in (2), /ne-/ precedes the fronted wh-phrase; note that here /ne-/ cannot also
occur pre-verbally:
(2)
a. *(N-)ke
(*n-)ge-it-ir-e
ege-toocho?
(foc-)what (foc-)sa-kill-prf-fv 7-rabbit
‘What killed the rabbits?’
b. *(N-)ke
ichi-ng’iti (*n-)cha-a-it-ir-e?
(foc-)what 10-hyena (foc-)10sa-pst-kill-prf-fv
‘What did the hyenas kill?’
Similarly, in (information) focus constructions, e.g. answers to wh-questions, as in
(3), /ne-/ appears on the fronted, focused phrase; here too, /ne-/ cannot also appear preverbally:
(3)
a. *(N-)ichi-ng’iti (*n-)cha-a-it-ir-e
ege-toocho.
(foc-)10-hyena (foc-)10sa-pst-kill-prf-fv 7-rabbit
‘THE HYENAS killed the rabbit.’
b. *(N-)ege-toocho ichi-ng’iti (*n-)cha-a-it-ir-e.
(foc-)7-rabbit 10-hyena (foc-)10sa-pst-kill-prf-fv
‘The hyenas killed THE RABBIT.’
Note that alongside (3a), the alternative word order in (4) is also possible, where the
subject precedes the constituent marked with /ne-/; as far as we can tell, there is no
difference in interpretation or contextual appropriateness between (3a) and (4).
(4) Ichi-ng’iti *(n-)ege-toocho (*n-)cha-a-it-ir-e.
10-hyena (foc-)7-rabbit (foc-)10sa-pst-kill-prf-fv
‘The hyenas killed THE RABBIT.’
It is not possible for /ne-/ to appear post-verbally, thus, attempting to focus an in-situ
object is ungrammatical:
(5) Ichi-ng’iti cha-a-it-ir-e
(*n-)ege-toocho.
10-hyena 10sa-pst-kill-prf-fv (foc-)7-rabbit
(Intended meaning: ‘The hyenas killed THE RABBIT.’)
To summarize this section, in declaratives, /ne-/ obligatorily appears pre-verbally,
while in wh-questions and focus constructions, /ne-/ obligatorily precedes the fronted
395
Meredith Landman & Rodrigo Ranero
constituent.6 We see the varied distribution of this morpheme as raising two questions
for a uniform account of its different uses. First, is /ne-/ truly a focus marker? Second,
where does /ne-/ originate syntactically? We put forward answers to these questions in
the next two sections. Specifically, in §3, we present evidence that /ne-/ semantically is
uniformly a focus marker, and in §4, we show that Schwarz’s (2007) syntactic account
of Kikuyu can be extended to Kuria, so that in all cases, /ne-/ heads a Focus Phrase projection in the left periphery of the clause.
3 Diagnosing focus
In this section, we present evidence that /ne-/ behaves like a focus marker, across its
different uses. Following the alternative semantics approach to focus (Rooth 1985; 1992),
we assume that ‘focus indicates the presence of alternatives that are relevant for the
interpretation of linguistic expressions’ (Krifka 2008). This definition encompasses different types of focus, e.g. signaling new information, correction, contrast, etc. We elicited
data from three types of contexts in which focus marking would be expected, and found
that /ne-/ consistently marks focused constituents, across syntactic categories. Specifically, following in part Hartmann & Zimmermann (2009)’s work on focus marking in
Gùrùntùm, we looked at the following four focus contexts in Kuria, all of which involve
morphological marking with /ne-/: (i) question-answer congruence, (ii) corrective focus,
(iii) contrastive focus, and (iv) association with focus sensitive operators such as Kuria
bene ‘only’. In the following subsections, we consider each type of focus context in turn.
3.1 Question-answer congruence
In felicitous answers to wh-questions, /ne-/ obligatorily appears on the phrase corresponding to the wh-phrase. Consider, e.g. the object wh-question in (6). A felicitous an6 A reviewer asks whether /ne-/ also appears in copular constructions, as in Kikuyu (Schwarz 2007). A nasal
morpheme does indeed appear in copular constructions, as in (ii) below, but investigating the distribution
of this morpheme and whether it is the same as /ne-/ goes beyond the scope of this paper:
(i)
Gati n-omo-reri.
1.Gati n-1-doctor
‘Gati is a doctor.’
Note also that the negative counterpart to /ne-/, /te-/, which we discuss in §6.2, also appears in copular
constructions:
(ii)
Gati t-omo-reri.
1.Gati t-1-doctor
‘Gati is not a doctor.’
396
24 Focus marking in Kuria
swer to this question is one where the object bears /ne-/, as in (7); answers in which /ne-/
appears on the subject, as in (8), are grammatical but infelicitous in this context.7
(6) Q: N-ke
ichi-ng’iti cha-a-it-ir-e?
foc-what 10-hyena 10sa-pst-kill-prf-fv
‘What did the hyenas kill?’
(7)
A1: N-ege-toocho ichi-ng’iti cha-a-it-ir-e.
foc-7-rabbit 10-hyena 10sa-pst-kill-prf-fv
‘The hyenas killed THE RABBIT.’
(8)
A2: #N-ichi-ng’iti cha-a-it-ir-e
ege-toocho.
foc-10-hyena 10sa-pst-kill-prf-fv 7-rabbit
(Intended meaning: ‘They hyenas killed THE RABBIT.’)
Consider also subject wh-questions and their answers, as in (7). A felicitous answer
to a subject wh-question exhibits a fronted subject DP bearing /ne-/, as in A1; if other
phrases bear /ne-/, the answer is infelicitous, as in (10).
(9) Q: Who ate mangoes?
A1: N-omo-onto a-a-rey-e
ama-yembe.
foc-1-person 1sa-pst-eat.prf-fv 6-mango
‘SOMEONE ate mangoes.’
(10) A2: #N-ama-yembe omo-onto a-a-rey-e.
foc-6-mango 1-person 1sa-pst-eat.prf-fv
(Intened meaning: ‘SOMEONE ate mangoes.’)
7
We have conflicting judgments from our speaker regarding whether pre-verbal /ne-/ as in (iv) below is
felicitous in contexts where we would expect the object to bear /ne-/. For example, (iv) below is sometimes
judged as infelicitous and sometimes as felicitous as an answer to (6); however, (7) is consistently offered
by our speaker as the first and best answer to the question in (6).
(i) A3: #Ichi-ng’iti n-cha-a-it-ir-e
ege-toocho.
10-hyena foc-10sa-pst-kill-prf-fv 7-rabbit
‘The hyenas killed the rabbit.’
397
Meredith Landman & Rodrigo Ranero
/Ne-/ marking in answers to wh-questions holds not only for object and subject whquestions, but also for adjuncts, such as PPs and adverbials. For example, (11) illustrates
a focused PP, and (12) illustrates a focused AdvP.8,9
(11)
Q: Where will Gati see the owl?
A: N-ko-mesa
gati umw-iti a-ra-maah-e.
foc-on-17.table 1.Gati 3-owl 3sg.sa-fut-see-fv
‘Gati will see the owl ON THE TABLE.’
(12)
Q: How did Chacha drink the chai?
A: M-bongo chacha a-a-nyoy-e
i-chaahe.
foc-quickly 1.Chacha 3sg.sa-pst-drink.prf-fv 8-chai
‘Chacha drank the chai QUICKLY.’
In answers to VP-oriented wh-questions, /ne-/ must occur pre-verbally, as the questionanswer pair in (13)-(14) shows; we take this to indicate that pre-verbal /ne-/ is also a focus
marker, in this case marking VP focus.10,11
Q: N-ke
gati a-a-korr-e.
foc-what 1.Gati 3sg.sa-pst-do.prf-fv
‘What did Gati do?’
(13)
8 PP
adjuncts canonically appear at the end of the sentence and manner adverbs are relatively free in their
positioning.
9
A reviewer asks whether in this case the PP S O V ordering is obligatory. We note that a postverbal object
is also possible, as in (v):
(i)
Q: Did you see the frog on the table or on the floor?
A: N-ko-mesa
naa-mah-er-e
i-kjoora.
foc-on-17.table 1sg.sa-pst-see-prf-fv 9-frog
‘I saw the frog ON THE TABLE.’
The PP S O V ordering in (8) above might be evidence for a TopP below FocP, although this matter needs
further investigation.
10
The example in (vi) below, in which the object bears /ne-/, is occasionally judged by our speaker as a
felicitous reply to the VP-oriented question in (13); thus, objects appear to project focus to VP in Kuria, as
has been observed for English (Selkirk 1984):
(i)
?N-i-chaahe a-a-nyoy-e.
foc-8-chai 3sg.sa-pst-drink.prf-fv
‘He DRANK CHAI.’
11
Note that while fronting the nominalized version of a verb is grammatical in Kuria, as in (vii) below, this
construction is infelicitous as an answer to a VP wh-question:
(i)
#N-oko-ria ama-ako b-a-rey-e.
foc-15-eat 6-fruit 3pl.sa-pst-eat.prf-fv
(Intended meaning: ‘They ATE FRUITS.’)
This contrasts with data reported for Kikuyu (see Schwarz 2007); see §5.2 below.
398
24 Focus marking in Kuria
(14)
A: N-a-a-nyoy-e
i-chaahe.
foc-3sg.sa-pst-drink.prf-fv 8-chai
‘He DRANK CHAI.’
Answers to questions in which sentential focus is expected, e.g. answers to questions
such as ‘What happened?’, also require pre-verbal /ne-/, as (15) shows; thus, preverbal
/ne-/ also marks sentential focus.12 Attaching /ne-/ to any other constituent, e.g. the
object, as in (16), would be infelicitous:
(15)
Q: What happened?
A1: Gati n-a-a-it-ir-e
ama-siisi.
Gati foc-3sg.sa-pst-kill-prf-fv 6-ant
‘Gati killed ants.’
(16) A2: #N-ama-siisi gati a-a-it-ir-e.
foc-6-ant 1.Gati 3sg.sa-pst-kill-prf-fv
(Intended meaning: ‘Gati killed ants.’)
Summarizing this subsection: (i) /ne-/ consistently marks focus expressions in answers
to wh-questions across categories; and (ii) pre-verbal /ne-/ marks VP focus as well as
sentential focus.
3.2 Corrective focus
Corrective focus contexts are those in which a (focused) phrase serves as a correction
to a like phrase already introduced into the discourse. Consider, e.g. the dialogue in (17)
below; in (18), speaker B corrects the VP from speaker A’s utterance. In this context,
/ne-/ may not occur on any phrase other than the corrected VP, as the infelicitous (19)
shows:13
(17) A: M-ba-a-gurr-i
i-nyamu.
foc-3pl.sa-pst-sell.prf-fv 9-cat
‘They sold the cat.’
(18) B: Aʔa, m-ba-a-gi-sirr-i.
no, foc-3pl.sa-pst-9om-lose.prf-fv
‘No, they LOST it.’
(19) C: #Aʔa, n-i-nyamu ba-a-sirr-i.
no, foc-9-cat 3pl.sa-pst-lose.prf-fv
(Intended meaning: ‘No, they LOST the cat.’)
12 A
reviewer notes that this instance of sentential focus looks identical to what we called a declarative in
(1). We wish to emphasize that this is a context where sentential focus would be expected, and so we take
/ne-/ to be indicating focus here.
13 A reviewer asks if something is focused in (17). We assume that the appearance of /ne-/ in this instance
shows sentential focus.
399
Meredith Landman & Rodrigo Ranero
Compare the dialogue in (20) and (21), where, in (21), B corrects the object DP from
A’s utterance in (20); in this case, the corrected DP object bears /ne-/:14
(20) A: Ichi-ng’iti n-cha-a-it-ir-e
ege-toocho.
10-hyena foc-10sa-pst-kill-prf-fv 7-rabbit
‘The hyenas killed the rabbit.’
B: Aʔa, n-in-chage cha-a-it-ir-e.
no, foc-9-zebra 10sa-pst-kill-prf-fv
‘No, they killed the ZEBRA.’
(21)
Corrective focus contexts thus provide further evidence that phrases bearing /ne-/ are
focused.
3.3 Contrastive focus
Contrastive focus contexts are those in which a phrase is presented in contrast with one
or more like phrases already introduced into the discourse. Consider for example (22),
where the contrastively focused VP in the conjoined clause bears /ne-/:15
i-indwi, na gati
(22) Chacha n-a-a-gorr-e
1.Chacha foc-3sg.sa-pst-buy.prf-fv 9-lion and 1.Gati
n-a-a-gurr-i
i-indwi.
foc-3sg.sa-pst-sell.prf-fv 9-lion
‘Chacha bought a lion and Gati SOLD a lion.’
Similarly, in (23), the contrastively focused DP in the conjoined clause bears /ne-/:16
i-nyamu, na gati n-i-indwi
Chacha n-a-a-gorr-e
and 1.Gati foc-9-lion
1.Chacha foc-3sg.sa-pst-buy.prf-fv 9-cat
a-a-gorr-e.
3sg.sa-pst-buy.prf-fv
‘Chacha bought a cat and Gati bought a LION.’
(23)
14 As
with question-answer congruence, we have conflicting judgments regarding whether pre-verbal /ne-/
is felicitous in object-focused contexts. Thus, (viii) below is occasionally judged as felicitous for corrective
focus on the object:
(i)
15 A
Aʔa, n-cha-a-it-ir-e
in-chage.
no, foc-10sa-pst-kill-prf-fv 9-zebra
‘No, they killed a ZEBRA.’
reviewer asks how we can tell that /ne-/ in the second conjunct in (22) marks VP focus. The alternative
would be to claim that nothing is marking VP focus in the second conjunct and it is just another case of
sentential focus (i.e., a declarative). Given that in this contrastive context we expect VP focus in the second
conjunct, we conclude that it is /ne-/ that marks focus.
16 A reviewer asks whether the appearance of /ne-/ in the first conjunct means that it is an instance of focus.
We assume that all sentences in this tense bear focus, and that the first conjunct is an instance of default
or sentential focus.
400
24 Focus marking in Kuria
Since /ne-/ appears on a contrastively focused phrase in these examples, they thus
also indicate that /ne-/ is a focus marker.
3.4 Focus sensitive operators
Finally, we consider focus sensitive operators analogous to English only, which have
been shown to associate with phrases bearing focus (see e.g. Rooth 1992; 1996). In Kuria,
focus sensitive operators such as bene ‘only’ associate with phrases bearing /ne-/. Consider e.g. (24) below, which is felicitous given the context:
(24)
Context: We are discussing the methods used by students to protest dining hall food
two weeks ago at school. There was only one method they used to protest and I
specify it as such.
M-ba-a-tan-er-a
g-oko-rekeera ama-geena bene.
foc-3pl.sa-pst-protest-prf-fv by-15-throw 6-stone
only
‘They only PROTESTED BY THROWING ROCKS.’
Bene may also associate with a focused DP, in which case bene displays noun class
agreement morphology, as evidenced by (25) (which is felicitous in a different context):17
(25)
N-i-nswi i-nyene ba-a-ta-rey-e.
foc-9-fish 9-only 3pl.sa-pst-neg-eat.prf-fv
‘They only didn’t eat FISH.’
That Kuria bene associates with phrases bearing /ne-/ also indicates that /ne-/ is a
focus marker.
3.5 Summary of focus diagnostics
Summarizing this section, we have shown that /ne-/ appears on just those phrases that
are in focus, based on four types of focus constructions: (i) question-answer congruence;
(ii) corrective focus; (iii) contrastive focus; and (iv) association of phrases bearing /ne-/
with focus sensitive operators.
4 Analysis
Having established that /ne-/ is semantically a focus marker, we turn now to its syntax.
Extending Schwarz (2007)’s account of Kikuyu to Kuria, we analyze /ne-/ as heading a
Focus Phrase (FocP) projection in the left periphery of CP (Rizzi 1997), as in (26).18
17
A reviewer asks whether this is the only reading for this sentence or whether any scope interactions exist;
this is indeed the only reading available for the sentence.
18
A TopP projection is observed in the tree, since we will argue for this position below.
401
Meredith Landman & Rodrigo Ranero
(26)
Position of ne
CP
C’
C
TopP
Top’
Top
FocP
Foc’
Foc
ne-
TP
Following Schwarz, we derive the different constructions that /ne-/ occurs in via movement of phrases bearing /ne-/ into [Spec, FocP].19 For example, in a sentence with a
fronted object, the object raises to [Spec, FocP], as (27) illustrates.20 Here, topicalization
of the subject is optional.
(27)
19
Object focus
[CP [TopP [FocP object1 [Foc’ [Foc ne- ] [TP …t 1 …] ] ] ] ]
We, like Schwarz (2007), are noncommittal with respect to how /ne-/ combines morphologically with the
constituent in [Spec, FocP]. Schwarz (2007: 144) notes two possibilities for Kikuyu ne: (i) ne itself heads FocP,
and cliticizes to whatever is in the specificer of FocP, and (ii) the focus feature is spelled out phonologically
as ne, and surfaces to the left edge of whatever occupies [Spec, FocP]. For concreteness, we adopt the first
possibility, and position /ne-/ as heading the focus phrase in the trees throughout.
20 A reviewer asks what the trigger for movement is in our analysis. While we do not spell out in detail
the mechanics of movement, our account is compatible with a Minimalist analysis (Chomsky 2000; 2004),
whereby A-bar movement is the result of an Agree operation between a head bearing an uninterpretable
feature (such as [wh], [Foc]) and a Goal which carries a matching feature. Furthermore, when the Probe
has the [EPP] property, this property is satisfied via internal merge of the Goal, which merges in the Spec
position of the Probe. While extending the Agree operation to account for A-bar movement in addition to
A movement is not uncontroversial—see the discussion in Horvath 2007 for instance—the present analysis
is compatible with accounts that take Agree and [Foc] features to drive syntactic movement.
402
24 Focus marking in Kuria
In the case of sentential focus, again following Schwarz (2007), TP moves up into
[Spec, FocP], while the subject is obligatorily topicalized, as in (28).21
(28)
Sentential focus
[CP [TopP subject2 [Top’ [FocP [TP1 …t 2 …] [Foc’ [Foc ne- ] [TP …t 1 …]]]]]
In the case of VP focus, VP moves to [Spec, FocP], while the subject is obligatorily
topicalized, again following Schwarz:
(29) VP focus
[CP [TopP subject2 [Top’ [FocP [VP1 …t 2 …] [Foc’ [Foc ne- ] [TP …t 1 …]]]]]]
Summarizing the analysis, we have extended Schwarz’s account of Kikuyu to Kuria,
so that in all of its uses, /ne-/ heads a Focus Phrase in the left periphery of the clause. We
turn first to a comparison of /ne-/ to morphological marking of focus in other languages
(§5), and then to further data regarding the distribution of /ne-/ (§6).22,23
21 A
reviewer asks whether movement of the subject out of the TP, which itself has moved to [Spec, FocP]
in (28) constitutes a violation of Rizzi 2010’s Criterial Freezing: “In a criterial configuration, the Criterial
Goal is frozen in place.” We believe that it does not, on the following grounds. Rizzi states that “In the
criterial configuration, only the element carrying the crucial feature is frozen in place, while the other
elements of the phrase pied-piped to the Spec of the Criterial Probe remain available for movement, and
can be subextracted, if no other syntactic principle is violated…” Although Rizzi does not specifically discuss
examples in which a TP moves to the specifier of a Criterial Probe, we hypothesize that in these cases the
subject is still available for movement, as observed in example (28).
22
An alternative account of /ne-/ might analyze it as a cleft construction. For example, Bergvall (1987) treats
Kikuyu /ne-/ as an assertion marker that heads TP, which would correctly account for the position of preverbal /ne-/. To account for clauses involving fronted phrases, Bergvall proposes a bi-clausal cleft analysis:
The matrix clause has a null expletive subject, and /ne-/ appears before a null copula, followed by a relative
clause. We see this type of account as problematic for Kuria on several counts. First, if these sentences are
clefts, we might expect that idiomatic readings should not be possible (e.g. in English It was the bucket that
he kicked lacks the idiomatic reading). However, idiomatic readings are preserved in Kuria, as the example
in (ix) illustrates:
(i) N-i-bara chonesi a-a-gey-e.
foc-9-wild 1.Johnes 3sa-pst-go.prf-fv
‘Johnes went INTO THE WILD.’ ~ ‘Johnes did something completely unexpected.’
Second, as Schwarz (2007) points out, clefts are not expected to occur in multiple wh-questions, such as
Who killed what? However, we do find such data in Kuria:
(ii) N-eng’we a-it-ir-e
(*n-)ke?
foc-who sa-kill-prf-fv foc-what
‘Who killed what?’
Third, /ne-/ appears in non-assertive contexts, e.g. yes/no questions, which differ tonally from declaratives but are otherwise identical (see Mwita 2008 for a discussion of Kuria tonology).
23
A reviewer asks whether an analysis along the lines of Horvath (2007), according to which an exhaustivity
operator merges with some phrase and then is attracted by a higher head, would also account for the Kuria
pattern. While the full consequences of extending Horvath’s analysis to Kuria is an issue for future research
(especially with regards to phrase-internal focus marking; see §5.1), we do not see how Horvath’s account
can be extended to Kuria for the following reason: Horvath justifies the exhaustivity operator based on
403
Meredith Landman & Rodrigo Ranero
5 Crosslinguistic comparison
In this section, we compare focus marking in Kuria to focus marking in other languages,
thereby placing the Kuria data in a crosslinguistic context.
5.1 Phrase-internal focus
In contrast with related Bantu languages (Abels & Muriungi 2008 on Kiitharaka) and
other language families (Hartmann & Zimmermann 2009 for Gùrùntùm), Kuria permits
phrase-internal focus marking. Consider first the following examples, which show that
/ne-/ can appear on the head noun in a fronted DP, (30), or on the determiner, (31):
(30)
Q: Which owls did Johnes see, these owls or those owls?
A: N-imi-iti ge-no chonesi a-a-roch-e.
foc-4-owl 4-this 1.Johnes 3sg.sa-pst-see.prf-fv
‘Johnes saw THESE OWLS.’
(31)
Q: Which owls did Johnes see, these owls or those owls?
A: Imi-iti n-ge-no chonesi a-a-roch-e.
4-owl foc-4-this 1.Johnes 3sg.sa-pst-see.prf-fv
‘Johnes saw THESE OWLS.’
In contrast with (30) and (31), if the determiner is kept constant and the nouns in the
two possible answers provided by the question differ, marking the determiner with /ne-/
becomes infelicitous:
(32)
Q: What did the children like, this lion or this rhinoceros?
A1: N-i-huuburia e-no b-a-tanch-er-e.
foc-9-rhino 9-this 3pl.sa-pst-like-prf-fv
‘They liked this RHINOCEROS.’
(33)
A2: #I-huuburia n-e-no
ba-a-tanch-er-e.
9-rhino
foc-9-this 3pl.sa-pst-like-prf-fv
(Intended meaning: ‘They liked this RHINOCEROS.’)
As a reply to questions that require a possessive DP as an answer, /ne-/ may also
appear phrase internally. Possessive DPs exhibit the following surface structure:
(34)
[Possessed DP Associative Marker Possessor DP]
the empirical observation that information focus remains in-situ in Hungarian, while contrastive/identificational focus undergoes overt movement. Since no such asymmetry exists in Kuria (i.e., both types of
focus are expressed in the same way, with /ne-/), we do not see that positing an exhaustivity operator is
justified for Kuria.
404
24 Focus marking in Kuria
The morpheme can attach to the possessed DP on the left edge of the phrase, or on
the associative marker. Consider first (35) below, where a question offers two alternative
answers in which the possessor DP is kept constant. Attaching /ne-/ to the left-edge of
the fronted phrase is the only felicitous reply in this context:
(35)
Q: What did Boke eat, Gati’s ugali or Gati’s chapati?
A1: N-iri-chabati re
gati a-a-rey-e.
foc-5-chapati 5.assoc 1.Gati 3sg.sa-pst-eat.prf-fv
‘He ate Gati’s CHAPATI.’
(36)
A2: #Iri-chabati ne-re
gati a-a-rey-e.
5-chapati foc-5.assoc 1.Gati 3sg.sa-pst-eat.prf-fv
(Intended meaning: ‘He ate Gati’s CHAPATI.’)
However, compare (35) with (37) below. Here, the question asks for the identity of
the possessor. As the second answer, (38), shows, only attaching /ne-/ to the associative
marker is felicitous in this context:
(37) Q: Whose ugali did Sammy eat?
gati a-a-rey-e.
A1: #N-ubu-kima bo
foc-14-ugali 14.assoc 1.Gati 3sg.sa-pst-eat.prf-fv
(Intended meaning: ‘He ate GATI’s ugali.’)
(38)
A2: Ubu-kima n-obo
gati a-a-rey-e.
14-ugali foc-14.assoc 1.Gati 3sg.sa-pst-eat.prf-fv
‘He ate GATI’S ugali.’
As a reply to the question in (37) above, we might have expected that marking the
possessor DP with /ne-/ would have been the only felicitous reply. However, it is ungrammatical (in any context) to mark the possessor DP with /ne-/, as (39) shows:24
(39) *Ubu-kima bo
n-gati
a-a-rey-e.
14-ugali 14.assoc foc-1.Gati 3sg.sa-pst-eat.prf-fv
(Intended meaning: ‘He ate GATI’s ugali.’)
Based on these examples, we wish to highlight the fact that /ne-/ may attach to different subconstituents of a focused phrase, depending on the context. Therefore, a potential
analysis treating the appearance of /ne-/ on different subconstituents of a focused phrase
to a semantically vacuous morphological operation would be undesirable. For now, we
leave the mechanics by which the focus marker appears phrase internally for future
research, although we can shed doubt on /ne-/ being base-generated phrase internally
given examples like (40) below, which show that marking /ne-/ on both the possessed
DP and the associative phrase is ungrammatical:
24 The ban on this type of construction was also noted for Kiitharaka by Abels & Muriungi (2008). We hypoth-
esize that the correct analysis of the surface structure of a possessive DP is one in which the associative
marker and the possessor DP form a single morphophonological unit that cannot be broken up by the
/ne-/ morpheme. Therefore, the ban against marking the possessor DP with /ne-/ might not be related to
semantics at all.
405
Meredith Landman & Rodrigo Ranero
(40) *N-ubu-kima n-obo
gati a-a-rey-e.
foc-14-ugali foc-14.assoc 1.Gati 3sg.sa-pst-eat.prf-fv
(Intended meaning: ‘He ate Gati’s ugali.’)
5.2 VP fronting
As reported by Schwarz (2007), VP focus in Kikuyu involves fronting an infinitival verb,
while a fully inflected form remains in base position (see his example 15B). This construction is also possible in Kuria, but does not trigger VP focus. Instead, fronting an
infinitival verb is felicitous only as a reply to a question expressing incredulity. Therefore, this construction can only be used to express a verum interpretation. Consider first
the following question-answer pairing, in which marking the VP with /ne-/ triggers VP
focus, as expected:
(41)
Q: What did they do, eat fruits or drink water?
A1: M-ba-a-rey-e
ama-ako.
foc-3pl.sa-pst-eat.prf-fv 6-fruit
‘They ATE FRUITS.’
(42) A2: #N-oko-ria ama-ako ba-a-rey-e.
foc-15-eat 6-fruit 3pl.sa-pst-eat-prf-fv
(Intended meaning: ‘They ATE FRUITS.’)
However, as a reply to the follow-up question observed below, only example (43) is
felicitous:
(43)
Q: Did they really?
A: E, n-oko-ria ama-ako ba-a-rey-e.
yes, foc-15-eat 6-fruit 3pl.sa-pst-eat.prf-fv
‘Yes, they DID eat fruits.’
These data indicate that in Kuria, verum focus in the remote past is marked via fronting
of an infinitival verb and predicate doubling. Further, these data show that Kuria differs
from Kikuyu with respect to the use of preverbal /ne-/ and the expression of verum focus:
In Kikuyu, preverbal /ne-/ is used for verum focus, while fronting the infinitival verb and
predicate doubling is akin to the use of preverbal /ne-/ in Kuria (what Schwarz 2007 calls
“narrow focus on the verb”). Although we leave an analysis of constructions such as (31)
for future research, we note the typological difference between Kikuyu and Kuria with
regards to the realization of VP and verum focus.25
25
The examples in (42) and (43) are parallel to data from other languages discussed in Aboh (2006) and Aboh
& Dyakanova (2009), where they are called “predicate fronting with doubling”. We refer the reader to
these papers for further examples of this type of construction crosslinguistically and thank an anonymous
reviewer for bringing these papers to our attention.
406
24 Focus marking in Kuria
6 Additional patterns regarding /ne-/
In this section, we document some additional patterns regarding the syntactic distribution of /ne-/, and in doing so articulate some puzzles for future research with respect to
focus marking in Kuria. Specifically, in §6.1, we observe certain contexts which appear
to involve focus, but lack /ne-/, and in §6.2, we discuss the complementary distribution
between /ne-/ and the negative marker /te-/.
6.1 Focus but no /ne-/
We have identified at least three contexts in which a phrase is semantically focused,
while /ne-/ is absent, which we discuss in the following subsections.
6.1.1 TAM
In certain TAM combinations, /ne-/ does not appear pre-verbally in certain contexts (as
also observed by Mwita 2008). Consider, e.g. the question-answer pairing in the immediate past below (cf. the remote past, in which /ne-/ does appear, as in, e.g. (1)):
(44) Q: What did Gati and Johnes just do?
A: Ba-rey-e
omo-gate.
3pl.sa-eat.prf-fv 3-bread
‘They just ATE BREAD.’
The morpheme may only appear pre-verbally in this tense as a reply to questions
showing incredulity, expressing a verum focus interpretation. This context, similar to the
previously discussed VP nominalization cases in the remote past, is exemplified below
in (45):
(45)
Follow up Q: Did they really?
A: M-ba-rey-e
omo-gate.
foc-3pl.sa-eat-prf.fv 3-bread
‘They DID just eat bread.’
We presently have no explanation for why /ne-/ only appears in verum focus contexts
in certain TAM combinations, as illustrated above, and why verum focus in certain TAM
combinations is not expressed via fronting of the infinitival verb and doubling.
6.1.2 Focus in relative clauses
Relative clauses provide another context in which /ne-/ does not appear pre-verbally.
Even in answer to an echo question context that forces a focus interpretation on the
verb, /ne-/ may not appear, as the question-answer pair in (46) shows:
407
Meredith Landman & Rodrigo Ranero
(46) Q: The teacher who did WHAT drank water?
A: Umw-arimu ora (*n-)a-a-giy-er-e.
1-teacher 1.who (foc-)3sg.sa-pst-march-prf-fv
‘The teacher who MARCHED.’
We might account for the restriction on /ne-/ in this context to the absence of a FocP
projection in the left periphery of relative clauses. However, it remains a puzzle how
phrases with no morphological marking are interpreted as being focused. We leave this
issue for future research.26
6.1.3 Multiple wh-phrases
Since we assume that congruent answers to wh-questions involve semantic focus, we
would expect all replies to wh-questions to bear /ne-/; this, however, is not the case.
Consider the multiple wh-question-answer pairing below; notice that only the phrase
that answers the wh-phrase marked with /ne-/ in the question can bear /ne-/ in the
answer; the in-situ constituent cannot also bear /ne-/:
(47)
Q: N-eng’we a-it-ir-e
(*n-)ke?
foc-who sa-kill-prf-fv (foc-)what
‘Who killed what?’
(48)
A1: M-boke
a-it-ir-e
igi-siisi.
foc-1.Boke 3sg.sa-kill-prf-fv 7-ant
‘Boke killed ANTS.’
(49)
A2: #N-igi-siisi boke a-it-ir-e.
foc-7-ant 1.Boke 3sg.sa-kill-prf-fv
(Intended meaning: ‘BOKE killed ANTS.’)
Examples like these again raise the question of how semantic focus is expressed and
interpreted in Kuria; if structural restrictions ban the appearance of multiple instances
of /ne-/ in a clause, how are phrases with no morphological marking interpreted as being
focused? We leave further discussion of this matter for future research.
6.2 The complementary distribution of /ne-/ and /te-/
While /ne-/ appears in positive sentences, its apparent negative counterpart, /te-/, appears in exactly the same position in negative sentences:
(50) Aba-saacha te-ba-a-mah-er-e
eng’-ombe.
2-man
neg.foc-3pl.sa-pst-see-prf-fv 9-cow
‘The men did not SEE THE COW.’
26 A reviewer notes that a fuller paradigm related to the unavailability of /ne-/ in relative clauses might show
this is due to island effects.
408
24 Focus marking in Kuria
The two are in complementary distribution, never co-occurring on the same phrase
or in the same clause. /Te-/ appears to express both negation as well as focus. Like /ne-/,
/te-/ attaches to focused phrases, such as the question-answer pairing below:
(51)
Q: Was it the men who saw the cow?
A: A:, *(t-)aba-saacha (*m-)ba-a-mah-er-e
eng’-ombe.
no, neg.foc-2-man (foc-)3pl.sa-pst-see-prf-fv 9-cow
‘It was not THE MEN who saw the cow.’
Also like /ne-/, focus sensitive semantic operators such as bene ‘only’ may associate
with /te-/. The example below is felicitous if the speaker is commenting on the methods
used by student protesters:
(52)
Te-ba-a-giy-ir-e
bene.
neg-foc-3pl.sa-pst-march-prf-fv only
‘They did not only MARCH.’
The complementary distribution of /ne-/ and /te-/ in Kuria suggests to us that the
two are closely related in their syntax and semantics, though we do not presently have
an explanation for this connection. Mwita (2008) attributes the incompatibility of /ne-/
with /te-/ to negation itself being “inherently focused”. However, this approach would
not account for the fact that /ne-/ can co-occur with a different negation marker, /ta-/,
which immediately precedes the verb root in contexts such as the following:27
(53)
Q: What won’t they do?
A: M-ba-taa-r-e
ege-eki.
foc-3pl.sa-neg.fut-eat-fv 5-cake
‘They will not EAT THE CAKE.’
A connection between the morphological marking of focus and negation has not, to
our knowledge, been studied in detail in the Bantu literature on focus realization. While
Schwarz (2007) and Abels & Muriungi (2008) acknowledge the incompatibility of the
focus morpheme with a negation morpheme in Kikuyu and Kiitharaka, respectively, they
do not propose an account of the pattern.28 We leave the issue for future research as well.
7 Conclusion
In this paper, we have presented a range of novel data regarding the syntax and semantics of the Kuria morpheme /ne-/. We have argued that our data support analyzing the
morpheme /ne-/ as a focus marker, and we have shown that an analysis in the spirit of
27
28
/Te-/ and /ta-/ may not both appear on the verb stem.
A similar connection between focus and negation manifests in English do-support, suggesting a crosslinguistic connection between focus and negation. E.g. in English, do-support is obligatory in negated
sentences (e.g. They *(did) not leave) as well as verum focus sentences (e.g. They *(DID) leave).
409
Meredith Landman & Rodrigo Ranero
Schwarz (2007) captures the syntactic distribution of /ne-/. This paper thus contributes to
our understanding of focus strategies in Bantu specifically, as well as across languages
more generally. Furthermore, we have articulated the following puzzles regarding the
realization of focus, not only for Kuria but across the Bantu family. First, we have documented patterns that involve phrase-internal focus marking, which have not previously
been reported in the Bantu literature. Second, we have shown how VP focus differs in
Kuria in comparison with other Bantu languages like Kikuyu. Finally, we have observed
the complementary distribution of focus and negation in Kuria, a connection that may
have implications for the analysis of focus in Kuria specifically, as well as focus across
languages more generally.
Acknowledgements
Authors are listed alphabetically. Special thanks to Johnes Kitololo for his judgments
and patience, to Michael Diercks for extensive input and advice, and to Adele Eslinger,
whose paper on Kuria /ne-/ for the Spring 2013 Field Methods course at Pomona College,
taught by Michael Diercks, provided a starting point for this work. Thanks also to Mary
Paster, Jesse Harris, the participants of the 2014 mini-symposium on African languages at
Pomona, the participants of ACAL 2014, and two anonymous reviewers for many helpful
questions and comments. This paper builds on the second author’s undergraduate thesis
at Pomona College. Any mistakes are our responsibility.
Abbreviations
Glosses are as follows:
foc
pst
prf
fut
fv
focus
past
perfective
future
final vowel
sa
neg
comp
assoc
appl
subject agreement
negation
complementizer
associative marker
applicative.
In the orthographic conventions used throughout, an intervocalic <b> represents a
voiced bilabial fricative [β], an intervocalic <g> a voiced velar fricative [ɣ], <ng’> a velar
nasal [ŋ], <ny> a palatal nasal [ɲ], <y> a palatal glide [j], <r> an alveolar tap [ɾ], <rr> a
voiced alveolar trill [r], and <ch> a voiced alveo-palatal affricate [ʧ]. Numbers indicate
Bantu noun class. For ease of comprehension, when one morpheme’s presence is to be
noticed in an utterance, it appears in bold; when two morphemes are to be noticed, the
second one is underlined. We do not transcribe tone in our data; see Mwita (2008) and
Marlo et al. (2014; 2015) for discussion of Kuria tone.
410
24 Focus marking in Kuria
References
Abels, Klaus & Peter Muriungi. 2008. The focus marker in Kiitharaka: Syntax and semantics. Lingua 118. 687–731.
Aboh, Enoch O. 2006. When verbal predicates go fronting. ZAS Papers in Linguistics 46.
21–48.
Aboh, Enoch O. & Marina Dyakanova. 2009. Predicate doubling and parallel chains. Lingua 119. 1035–1065.
Bergvall, Victoria Lee. 1987. Focus in Kikuyu and Universal Grammar. Cambridge: Harvard University Doctoral dissertation.
Cammenga, Jelle. 2004. Igikuria phonology and morphology: A Bantu language of southwest Kenya and northwest Tanzania. Köln: Rudiger Koppe Verlag.
Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Roger Martin, David
Michaels & Juan Uriagereka (eds.), Step by step: Essays on minimalist syntax in honor
of Howard Lasnik, 89–156. Cambridge, MA: MIT Press.
Chomsky, Noam. 2004. Beyond explanatory adequacy. In Adriana Belletti (ed.), Structures
and beyond: The cartography of syntactic structures, vol. 3, 104–131. New York: Oxford
Univerity Press.
Eslinger, Adele. 2013. The mystery Nasal. Ms. Pomona College.
Hartmann, Katharina & Malte Zimmermann. 2009. Morphological focus marking in
Gùrùntùm (West Chadic).
Horvath, Julia. 2007. Separating ‘focus movement’ from focus. In Simin Karimi, Vida
Samiian & Wendy Wilkins (eds.), Phrasal and clausal architecture, 108–145. Amsterdam:
John Benjamins.
Krifka, Manfred. 2008. Basic notions of information structure. Acta Linguistica Hungarica
55. 243–276.
Marlo, Michael R., Leonard Chacha Mwita & Mary Paster. 2014. Kuria tone melodies.
African Linguistica XX. 277–294.
Marlo, Michael R., Leonard Chacha Mwita & Mary Paster. 2015. Problems in Kuria H
tone assignment. Natural Language and Linguistic Theory 33. 251–265.
Mwita, Chacha. 2008. Verbal tone in Kuria. Los Angeles: University of California Doctoral
dissertation.
Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of grammar: A handbook in generative syntax, 281–337. Kluwer: Dordrecht.
Rizzi, Luigi. 2010. On some properties of Criterial Freezing. In Phoevos Panagiotidis (ed.),
The complementizer phase: subjects and operators, 17–32. Oxford: Oxford University
Press.
Rooth, Mats. 1985. Association with focus. Amherst: University of Massachusetts Doctoral
dissertation.
Rooth, Mats. 1992. A theory of focus interpretation. Natural Language Semantics 1. 75–
116.
Rooth, Mats. 1996. Focus. In Shalom Lappin (ed.), The handbook of contemporary semantic
theory, 271–297. London: Basil Blackwell.
411
Meredith Landman & Rodrigo Ranero
Schwarz, Florian. 2007. Ex-situ focus in Kikuyu. Focus Strategies in African Languages:
The Interaction of Focus, Grammar in Niger-Congo & Afro-Asiatic. Walter de Gruyter.
Selkirk, Elisabeth. 1984. Phonology and syntax. Cambridge, Massachusetts: MIT Press.
412
Chapter 25
A corpus study of the Swahili
demonstrative position
Mohamed Mwamzandi
Synchronic studies on Swahili adnominal demonstratives have not addressed the interplay
between syntactic position and pragmatic function of these structures. This study shows
how referential givenness of discourse entities may explain Swahili word order variation in
Swahili adnominal demonstratives. Class 1 (animate nouns) demonstratives are examined in
the two attested word orders: NP+DEM and DEM+NP. A close analysis of dataset extracted
from the Helsinki Corpus of Swahili reveals that the two structures have distinct pragmatic
values. The NP+DEM order is used for active topics while the DEM+NP order reactivates
semiactive/inactive topics. This study reveals how the syntax-pragmatics interplay may explain distinct structures viewed as semantic equivalents by native speakers.
1 Introduction
This paper exploresword order variation in Swahili adnominal demonstratives via corpus analysis. The term “adnominal demonstrative” is used in the literature to distinguish
demonstratives that co-occur with nouns from stand-alone pronominal demonstratives.
While an adnominal demonstrative forms a constituent with an adjacent noun, a pronominal demonstrative is a noun phrase in its own right. More specifically, I analyze the pragmatic use of Swahili demonstratives as outlined by Fillmore (1975; 1982; 1997). Thereafter,
I present a qualitative and quantitative analysis of the pre and postnominal position of
the Swahili demonstrative. I focus on the relationship that exists between cognitive level
of the hearer on discourse entities and the choice of referring expressions (Chafe 1987;
Ariel 1988; 1991; 2001; Gundel et al. 1993).
Swahili has various proximal and distal demonstrative forms that obligatorily agree
with the nominal class of the noun they modify as exemplified in Table 1.
Notice that the hV- stem is used for the proximal demonstrative while the -le stem is
used for the distal demonstratives. Further, the agreement affix varies with noun class
hence yu- and wa- for class 1 and 2 and u- and i- for class 3 and 4.
Besides the semantic distinction of distal and proximal demonstratives, there are two
demonstrative constructions that vary in their word order: NP + DEM as seen in (1) and
DEM + NP as seen in (2).
Mohamed Mwamzandi. A corpus study of the Swahili demonstrative position. In Jason Kandybowicz, Travis Major, Harold Torrence & Philip T. Duncan (eds.), African linguistics on the prairie:
Selected papers from the 45th Annual Conference on African Linguistics, 413–429. Berlin: Language
Science Press. DOI:10.5281/zenodo.1251756
Mohamed Mwamzandi
Table 1: Proximal and distal demonstrative forms of the first four noun classes.
Noun class
Proximal Dem
Distal Dem
1
2
3
4
hu-yu
ha-wa
hu-u
hi-i
yu-le
wa-le
u-le
i-le
(1)
[Msichana yule]
a-li-ingia.
1.girl
1.dist.dem 1.sm-pst-enter
‘That girl entered.’
(2)
[Yule
msichana] a-li-ingia.
1.dist.dem 1.girl
1.sm-pst-enter
‘That girl entered.’
The distal demonstrative yule ‘that’ is postnominal in (1) but prenominal in (2). The
general tendency in studies on the demonstrative position in Bantu is to claim that the
postnominal demonstrative (1) is the unmarked form reserved for the basic gestural function, while the prenominal demonstrative (2) is an innovation aimed at marking definite
reference (Ashton 1944; Carstens 1991; 2008; Tamanji 2006). Amidu (2006) points out
that both the pre and postnominal demonstrative positions as seen in (1) and (2) can
be referential (anaphoric due to previous mention) but does not discuss the pragmatic
implications of these demonstratives.
In my analysis, I first discern the adnominal demonstrative function as gestural, anaphoric, or recognitional (Fillmore 1975; 1982; 1997; Himmelmann 1996; Diessel 1999).
Thereafter, I qualitatively and quantitatively analyze the pragmatic function of the Swahili demonstrative position. I posit that the postnominal demonstrative as seen in (1) indicates that the intended referent is “active” (Chafe 1987). On the other hand the prenominal demonstrative as seen in (2) indicates that the intended referent is “semi-active”
or “inactive”. “Semiactive” referents are those discourse entities reintroduced in the discourse after topic shift (change of topic) as well as discourse entities within the conversational context. Topic in this study is what an utterance is about. “Inactive” topics are
(re)introduced in the discourse after a long gap of absence or are familiar to the interlocutors.
It is important to note here that Chafe’s (1987) activation states as outlined above
do not make specific claims on the relationship that exists between activation level
and forms of referring expressions. To tackle this absence of matching activation level
with forms of referring expression, I invoke the Accessibility Hierarchy (Ariel 1988; 1991;
2001) and the Givenness Hierarchy (Gundel et al. 1993) cognitive theories which associate referential choice with “referential givenness”: The awareness level of interlocutors
to discourse entities (Gundel & Fretheim 2006). These two cognitive hierarchies rank
414
25 A corpus study of the Swahili demonstrative position
demonstrative expressions as mid-accessibility markers. Pronouns are ranked higher
than demonstratives while explicit NPs are ranked lower.
A few things on the scope and limitations of this study are worth mentioning. In this
study, I examine class 1 (animate nouns) proximal (hu-yu) and distal (yu-le) demonstratives. Class 1 is chosen because of the relative prominence and sustainability of animate
nouns as opposed to inanimate nouns in discourse (Givón 1976; 1983). The applicability
of the results is therefore limited to class 1 demonstratives though an extension of the
findings to other noun classes is plausible. Further, this study does not look at the distribution of referential demonstratives. Referential demonstratives such as huyo are formed
by suffixing the “O” of reference to the proximal demonstrative and then deleting the
final vowel of the demonstrative (Ashton 1944). While the referential demonstrative is
mainly used in discourse to mark definiteness, the use of a proximal/distal demonstrative
is not limited to this function (See §2.2.1). Due to its difference in form and functional
limitation, the distribution of the referential demonstrative is left out for future research.
The organization of the rest of the paper is as follows. In §2 I explain the methodology. In §3 I present and discuss the results of the study. §4 presents the conclusion and
theoretical implications.
2 Methodology
In this section, I explain extraction of the dataset from the Helsinki Corpus of Swahili in
§2.1. I then discuss how the dataset was coded in §2.2.
2.1 Extraction of the dataset from the corpus
The source of data in this study is the Helsinki Corpus of Swahili (HCS) which has 14
annotated corpora. The corpora contain current newspaper articles as well as excerpts of
literary texts, education and science material written in the mid to late 20th century. Due
to the absence of annotations on anaphora resolution in the corpora, I limit the analysis
to the HCS books (cf. Mitkov 1994). The HCS books sub-corpus has 1,055,425 words in 71
documents. The documents are mainly Swahili literary texts and education manuscripts.
To obtain the dataset, four queries were made in the HCS. Due to limitations associated with functionality of corpus software, the queries asked for all nouns adjacent to
demonstratives whether the demonstrative and the adjacent noun formed a syntactic
unit or not. Thus, a manual postediting process aimed at eliminating all the DEM+NP
collocations that did not form a syntactic unit was done. Most of these cases were ditransitive verbs with a demonstrative adjacent to both the direct object and the oblique
argument as seen in (3).
(3)
Njoo u-m-pat-i-e
[kijana huyu]
[maji ya kunywa].
come 2sg-om-get-appl-imp teenager prox.dem water of drinking
‘Come and give this teenager some water to drink.’
415
Mohamed Mwamzandi
In (3), the proximal demonstrative huyu is modifying the direct object kijana ‘teenager’
but was also displayed by the HCS concordancer as a prenominal demonstrative modifying the indirect object maji ya kunywa ‘drinking water’. Other cases that were eliminated
include pronominal identification demonstratives in which the copula introducing the
demonstratum was deleted; and adnominal demonstratives from poems whose pre or
postnominal position may be driven by metrical requirements. Table 2 shows the number of adnominal demonstratives before and after disambiguation.
Table 2: Adnominal demonstratives before and after postediting.
And-Dem
And-Dems before
disambiguation
And-Dems after
disambiguation
Difference
133
135
140
114
522
109
124
126
75
434
24
11
14
39
88
Prenominal proximal
Postnominal proximal
Prenominal distal
Postnominal distal
Total
Each of the disambiguated demonstrative expression was then displayed in its narrow
context (in the HCS of Swahili concordancer) as well as its wider context (in the original
text) for contextual and statistical analysis.
2.2 Coding the data
Each demonstrative expression was coded for the following variables: dem-type (proximal, distal), dem-function (gestural, anaphoric, recognitional), dem-position (prenominal, postnominal) and the activation state (active, semiactive, inactive). Anaphoric demonstratives were further coded for referential distance. While coding for dem-type and demposition was straightforward after displaying the queries in their wider context, coding
for the dem-function, referential distance and activation state needs further elaboration.
Each of these variables is explained in turn in §2.2.1, §2.2.2 and §2.2.3.
2.2.1 Demonstrative function
Adnominal demonstratives as referring expressions have mostly been analyzed by looking at the demonstrative function: gestural, anaphoric, and recognitional. Coding for
these demonstrative functions is explained below.
‘Gestural’ here does not necessarily mean actual pointing but rather situations which
need ‘pointing’ of some sort to establish reference. In the dataset there are instances such
as (4) where a cue word may indicate that the demonstrative in question is gestural.
416
25 A corpus study of the Swahili demonstrative position
(4) [Yule
bwana] u-na-mu-on-a:
Mzalamo yule?
dist.dem person 2sg-prs-om-see-fv Zaramo dist.dem
‘Do you see that person: is he a Zamoro (ethnic community)?’
In (4), the demonstrative expression yule bwana ‘that person’ was coded gestural because the verb on-a ‘see’ draws the attention of the hearer to a potential discourse entity
within the conversational context. Only first mentions of referents within conversational
contexts were coded as gestural. Subsequent mentions were coded as anaphoric.
Anaphoric demonstratives track discourse entities across clauses (intra-sentential) (5)
as well as across sentences (inter-sentential) (6) (Botley & McEnery 2000).
(5) A-li-po-fik-a
kwa [mzee Malongo], [mzee yule]
sm-pst-when-arrive-fv at old.man Malongo old.man dist.dem
a-ka-shangaa.
sm-seq-surprise
‘When he (Kiliilo) arrived at mzee Malongo’s home, that old man (Malongo) was
surprised (to see him).’
(6)
a. U-ki-vuka
bahari saba, ku-na [chewa]i mkubwa.
2sg-cond-cross seas seven, 17sm-be grouper big
‘If you cross the seven seas, there is a grouper (type of fish).’
b. [Chewa huyu]i a-ki-vuta
pumzi
1grouper this 1sm-cond-breath air
‘When this grouper is breathing…’
In (5), the NP mzee Malongo in the matrix clause is the antecedent of the demonstrative expression mzee yule ‘that mzee’ in the embedded clause. In (6a), the noun chewa
‘grouper’ is the antecedent of the demonstrative NP chewa huyu ‘this grouper’ in (6b).
Demonstratives used to track referents in intra and intersentential contexts were coded
as anaphoric.
Demonstratives used recognitionally indicate common knowledge and therefore do
not have a co-specification element in the surrounding situation or preceding discourse
(Diessel 1999). This is illustrated in (7).
(7)
[Huyu
Juma] ka-shindw-a
ku-ku-tunz-a.
prox.dem Juma sm.prf-defeat-fv inf-om-take.care-fv
‘This Juma has failed to provide for you.’
In (7) the proximal demonstrative huyu indicates that Juma is the man the speaker and
the hearer all know. The demonstrative expression here is not anaphoric since the referring expression Juma has no apparent antecedent in the preceding discourse. It is also
not gestural because the referent Juma is not physically present in the conversational
context.
417
Mohamed Mwamzandi
Although recognitional demonstrative expressions are overwhelmingly used in first
mentions to indicate common knowledge, there are instances where subsequent mentions via a demonstrative expression may mark the referent as familiar at that point of
discourse. This is illustrated in (8).
(8)
Kumbe [yule
mtu mweupe] amba-ye a-li-kuwa a-me-nusur-ik-a
intj dist.dem man white
comp-rel sm-aux sm-perf-save-stv-fv
ku-ua-w-a
na wenyeji
inf-kill-pass-fv by natives
‘Alas, that white man who had escaped being killed by the natives … ’
In (8), the referential distance between the adnominal demonstrative and its antecedent
was 118 clauses. The writer is aware of the “referential problem” (Auer 1984) caused by
topic shift and therefore adds more information to the adnominal demonstrative in form
of a restrictive clause to ensure successful identification of the referent. Following Himmelmann (2006: 230), in addition to first mention of discourse entities to indicate common knowledge, I also coded demonstrative expressions as recognitional if the gap of
absence after previous mention was too long to warrant “additional anchoring or descriptive information to make the intended referent more accessible”.
2.2.2 Referential discourse
Referential distance has been described as the most important diagnostic tool for measuring referential givenness. Givón (1983: 36), for example, explains that the effect of
referential givenness on accessibility correlates with other factors such as interference
from other possible discourse entities since “a high referential distance would show - all
other things being equal - more interfering topics in the preceding register.” Interfering
topics are other topics mentioned other than the immediate topic before its previous
mention in the discourse.
Since the finite clause is the locus for topic update, referential distance in this study
is the number of finite clauses from the relevant adnominal demonstrative expression
to a co-specifying explicit NP to its left (cf. Kameyama 1998; Poesio et al. 2004; Taboada
& Zabala 2008). The clause as the ‘locus for topic update’ implies that it is at the clause
level that more information about the topic is added. Example (9) illustrates how coding
for referential distance with the finite clause as the unit of analysis was done. Notice that
each of the clauses in (9a-c) contains new information about the topic (mjumbe ‘messenger’). The letter u stands for ‘utterance’ – the minimal unit of analysis in discourse, in
this case, the finite clause (cf. Grosz et al. 1995).
(9)
418
a. (u1) mjumbe wa tano alipotakikana, (u2) alitokea bila ya ajizi. (u3) Mjumbe
huyu alikuwa Ridhaa
[Mjumbe] wa tano a-li-po-tak-ik-an-a,
messenger of fifth sm-pst-when-want-stv-recp-fv
When the call for the fifth messenger was made,
25 A corpus study of the Swahili demonstrative position
b. a-li-toke-a
bila
ya ajizi
sm-pst-appear-fv without of fail
he came forth without fail.
c. [Mjumbe huyu]
a-li-kuwa Ridhaa.
messenger prox.dem sm-pst-aux Ridhaa
This messenger was Ridhaa.
The ref-distance in (9) was coded as 2, that is, there are two finite clauses before the
subsequent mention of the topic mjumbe in (9c).
2.2.3 Activation states
Depending on the referential distance between the adnominal demonstrative under consideration and its antecedent, the adnominal demonstrative in question was coded as active, semi-active or inactive. A question that arises under this description adapted from
Chafe (1987) is: What is the number of intervening utterances that qualify a discourse
entity to be active/semiactive/inactive?
The intended referent of an active referent is within the immediate consciousness of
the discourse participants. Thus, an adnominal demonstrative expression was coded as
‘active’ if there was an apparent antecedent in the preceding utterance as is the case in
(10).
(10)
a. Mtu wa pili
ku-kut-an-a
na-ye
a-li-kuwa [mzee].
person of second inf-meet-recp-fv with-3sg sm-pst-aux old.man
‘The second person to meet me was an old man.’
b. [Mzee huyu]
a-li-kuwa a-ki-peleka ng’ombe wake mtoni.
old.man prox.dem sm-pst-aux sm-ipfv-take cows
his river-loc
‘This old man was taking his cows to the river.’
In (10a) the NP mzee ‘old man’ is an apparent antecedent of the adnominal expression
mzee huyu ‘this old man’ in (10b). The adnominal demonstrative expression mzee huyu
‘this old man’ in (10b) was therefore coded as active.
Semiactive referents in this study were of two types: situational (in conversational
context) and textual (in discourse texts). Consequently, all gestural adnominal demonstratives were coded as semiactive. In discourse texts, a referent was coded as semiactive if there was an intervening topic(s) between the previous explicit mention of the
antecedent NP to the adnominal demonstrative expression under consideration. This is
illustrated in (11) and (12).
(11) Mbele ya-ngu ku-li-kuwa
bado watu wawili [yule
mzee] na
in.front poss-1sg loc17-pst-aux still people two dist.dem old.man and
[msichana mmoja].
girl
one
‘In front of me, there were still two people remaining, that old man and one girl.’
419
Mohamed Mwamzandi
(12)
[Msichana huyu],
a-li-ye-kuwa
bado a-me-weka kitambaa
girl
prox.dem sm-pst-rel-aux still sm-prf-put handkercheif
‘This girl, who still had a handkercheif placed … ’
In (11), yule mzee ‘that old man’ and msichana mmoja ‘a girl’ are the potential topics
for the following utterance. A potential topic is a referent within an utterance that can be
chosen by the speaker to be the center (topic) of the next utterance (cf. Grosz et al. 1995).
In the following 4 utterances (not presented above), the mzee ‘old man’ is established and
continued as the topic. In (12), the demonstrative expression msichana huyu ‘this girl’
reintroduces the girl mentioned in (11). The adnominal demonstrative msichana huyu
‘this girl’ in (12) was therefore coded as semiactive because of the interfering topic, mzee
‘old man’.
All recognitional demonstratives were coded as “inactive” because their identification
depends on retrieval of the discourse participants from the memory (see §2.2.1).
3 Results and discussion
In this section, I discuss the relevance of the demonstrative function in explaining the
demonstrative position in §3.1. I then discuss the relationship that exists between the
demonstrative position and activation states (active, semiactive, inactive) in §3.2.
3.1 Demonstrative function and position
Of the 434 adnominal demonstratives in the dataset, gestural demonstratives were 52,
anaphoric 308 and recognitional 74. The frequencies of dem-type (proximal and distal)
in both the pre and postnominal position are presented in Table 3.
Table 3: Dem-function and dem-position in proximal and distal demonstratives.
Gestural
Proximal
Distal
Anaphoric
Recognitional
Pre
Post
Total
Pre
Post
Total
Pre
Post
Total
38
2
9
3
47
5
49
83
110
66
159
149
22
41
5
6
27
47
The pragmatic value of the demonstrative position for each of the demonstrative functions will be discussed in turn.
3.1.1 Gestural demonstratives
Table 3 above shows that the proximal gestural demonstratives in prenominal position
were 38 and 9 in postnominal. There were 2 distal gestural demonstratives in prenominal
420
25 A corpus study of the Swahili demonstrative position
position and 3 in postnominal. These frequencies show that, first, the proximal demonstrative is mostly used as the deictic expression for the gestural function. The total frequency of proximal gestural demonstratives is 47 while the total frequency of the distal
gestural demonstratives is 5. This frequency difference is significant (X2 (1,N=52)=33.92,
p < 0.001). Second, the gestural demonstratives have a higher frequency count in prenominal position than postnominal. The total number of prenominal demonstratives is 40
while in the postnominal position the total number is 12. This frequency difference is
also significant (X2 (1,N=52)=15.08, p < 0.001).
The difference in the demonstrative position for the gestural demonstratives can be
explained by recalling the grammaticalization of the Swahili prenominal demonstrative
to express definite reference (Ashton 1944; Givón 1976; Carstens 1991; 2008). In their
paper on definite reference in English, Clark & Marshall (1981: 38), mention physical
copresence (presence in conversational contexts) as one of the reasons which license
definite reference in English. Based on the contextual analysis of the corpus data, I posit
that the prenominal demonstratives are mostly used to point to definite referents due to
physical copresence as seen in (4) repeated here as (13).
(13) [Yule
bwana] u-na-mu-on-a:
Mzalamo yule?
dist.dem person 2sg-prs-om-see-fv Zaramo dist.dem
‘Do you see that person: is he a Zamoro (ethnic community)?’
Based on the high frequency of gestural demonstratives in prenominal position, it
can be deduced that the prenominal position is mainly used to mark the referents as
accessible (semi-active) in conversational contexts. The examples in (14) and (15) further
illustrate this.
(14)
Huyu
kondoo tu-m-pelek-e
kwa Mfalme Ndevu.
prox.dem sheep 1pl-om-take-imp to King Ndevu
‘This sheep, let us take her to King Ndevu.’
(15)
Mfalme a-ki-m-pat-a
kondoo huyu
a-ta-furahi
sana.
King sm-sbjv-om-get-fv sheep prox.dem sm-fut-happy very
‘If the King gets this sheep, he will be very happy.’
In (14), because the discourse participants are all aware of the sheep’s presence within
the conversational context, the prenominal demonstrative in huyu kondoo ‘this sheep’
marks the referent as definite due to physical copresence. In (15), however, the postnominal position of the demonstrative marks the previously mentioned kondoo ‘sheep’
as anaphoric. As it will be seen in §3.1.2, the unmarked position for anaphoric demonstratives is postnominal.
3.1.2 Anaphoric demonstratives
The distribution of the 308 anaphoric demonstratives was as follows. There were 49
proximal demonstratives in prenominal position but 110 in postnominal position (X2
421
Mohamed Mwamzandi
(1,N=159)=23.40, p<0.001). The distal demonstratives were 83 in prenominal position but
66 in postnominal position, p>0.05. These results show that for the anaphoric demonstratives the proximal demonstrative has a higher frequency in postnominal position than in
prenominal. When contrasted with the distal postnominal demonstrative, the proximal
postnominal demonstrative frequency is also significantly higher (X2 (1,N=176)=11.00,
p<0.001). In the prenominal position, the distal demonstrative has a significantly higher
frequency than the proximal demonstrative (X2 (1,N=132)=8.76, p<0.005).
In order to further explore the frequency tendencies of the anaphoric demonstratives,
the referential distance of the anaphoric demonstrative expressions in the dataset was
analyzed. The results are presented in §3.1.3.
3.1.3 The effect of referential distance on the anaphoric demonstrative position
In measuring the referential distance, the number of finite clauses between an adnominal
demonstrative and a co-referential NP to its left was counted and recorded in a database.
The raw data was then log transformed to reduce the skewness of the data distribution.
After log-transformation, the Shapiro-Wilk test revealed that the data distribution for
the distal and proximal prenominal demonstratives was normal, p>0.05. The skewness
of the distal and proximal postnominal demonstrative data was greatly reduced but not
completely eliminated, p<0.05.1
Table 4 and Table 5 report the descriptive statistics of the demonstrative position for
the raw and log-transformed data respectively. The number in parentheses is the standard deviation while the number outside the parentheses is the mean referential distance.
Table 4: Mean referential distance and standard deviation of raw data.
Dem_Type
Proximal
Distal
Prenominal
Postnominal
5.55 (5.39)
7.40 (6.55)
5.25 (5.06)
5.29 (4.35)
Table 5: Mean referential distance and standard deviation of log-transformed
data.
Dem_Type
Proximal
Distal
1
Prenominal
Postnominal
1.34 (0.87)
1.70 (0.77)
1.25 (0.92)
1.30 (0.89)
The statistical operations conducted in this study assume normal distribution. Log transformation of the
variables enhances normal distribution, hence reducing the influence of outliers on the results (Baayen
2008).
422
25 A corpus study of the Swahili demonstrative position
The mean referential distance of the prenominal demonstratives is higher than that of
postnominal demonstratives. A non-repeated measures ANOVA with ref-distance as the
dependent variable and dem-type and dem-position as the independent variables reveal
that there is a significant main effect of ref-distance on dem-type (F(1,308)=6.09, p<0.05)
and dem-position (F(1,308)=5.90, p<0.05). There was no significant interaction between
dem-type and dem-position, p>0.05.
Further, a planned comparison using the t.test reveals that the mean ref-distance of
the distal prenominal demonstrative is higher that of the proximal prenominal demonstrative, p<0.05. The nonparametric Wilcoxon test applied to compare the median of
the distal prenominal demonstrative and the distal postnominal demonstrative indicates
that the medians of these two vectors and their distributions are different. Hence, the
mean referential distance of the distal prenominal demonstrative is also higher than that
of the postnominal distal demonstratives, p<0.05. However, there is no significant difference between the mean ref-distance of the proximal pre and postnominal demonstratives.
These statistics show that:
1. The difference in referential distance between the proximal and distal postnominal
demonstrative is not significant.
2. The distal prenominal demonstrative tends to be separated from its antecedent by
longer referential distance than the distal postnominal demonstrative as well as
the proximal pre and postnominal demonstrative.
3. The difference in referential distance between the proximal pre and postnominal
demonstratives is not significant.
I illustrate these observations with examples from the corpus.
These statistics show that the proximal demonstrative is frequently used in postnominal position when the referential distance is short (See example (15)). The insignificant
difference in referential distance between the proximal and distal postnominal demonstratives further suggests that there are cases when a distal postnominal demonstrative
may be used after a short referential distance as seen in (16).
(16)
ku-ingia ndani,
a. Adili a-li-po-taka
Adili sm-pst-when-want inf-enter inside
‘When Adili was about to go inside (the house), ’
b. a-li-ona [mtu] a-me-simama mlango-ni
sm-pst-see person sm-prf-stand door-loc
‘he saw a person standing at the door … ’
c. Adili a-li-dhani
[mtu yule]
a-li-kuwa bawabu.
Adili sm-pst-assume person dist.dem sm-pst-aux security.officer
‘Adili thought that the person was a security officer.’
423
Mohamed Mwamzandi
The referent mtu ‘person’ introduced in (16b) is continued in (16c). The postnominal
position of the demonstrative in (16c) marks the referent as ‘active’. The use of the postnominal distal demonstrative yule ‘that’ instead of the proximal demonstrative huyu
‘this’ has a special effect of marking the “narrative distance” (Leonardo 1987; Wilt 1987),
that is, the author is narrating events from a third person’s perspective. In the third
person’s perspective style of narration, the narrator is not involved in the events of the
story.
Further, the results show that the distal prenominal demonstrative is separated from
its antecedent by long referential distance as illustrated in (17).
(17)
[Yule
msichana] a-li-ingi-a.
dist.dem girl
sm-pst-enter-fv
‘That girl entered.’
In (17), the demonstrative expression yule msichana reintroduces the girl as the topic
after 45 clauses.
It is important to mention here that most corpus generalizations are based on statistical tendencies (See Mwamzandi 2014 for more examples). In general, anaphoric proximal and distal demonstrative are used postnominally after a short referential distance to
mark the intended referent as active. Anaphoric distal demonstrative are used prenominally after topic shift to mark the referent as semiactive.
3.1.4 Recognitional demonstratives
The frequency of recognitional proximal demonstratives in prenominal position was
22, and 5 in postnominal position (X2 (1,N=27)=10.70, p < 0.01). In prenominal position, the frequency of distal demonstratives was 41, and 6 in postnominal position (X2
(1,N=47)=26.06, p < 0.001). The difference between the recognitional demonstratives in
pre and postnominal positions is statistically significant (X2 (1,N=74)=36.54, p < 0.001). It
can be inferred from the results that a demonstrative is preferred in prenominal position
if used recognitionally.
Contrary to Himmelmann’s (1996) claim that only one of the demonstratives, mostly
the distal demonstrative, is preserved for the recognitional function across languages,
both the distal and proximal demonstratives can be used for this function in Swahili as
seen in (18) and (19).
(18)
[Yule
mtoto wako] a-na-ye-fundisha Chuo Kikuu,
dist.dem child your sm-prs-rel-teach university
‘That child of yours who teaches at the university, …’
(19) Hii
ni kazi ya majirani zetu, hasa
[huyu
mjukuu wa
9prox.dem is 9work of neighbours our especially prox.dem grandchild of
Ndenda].
Ndanda
‘This is the work of our neighbours, especially this grandchild of Ndenda.’
424
25 A corpus study of the Swahili demonstrative position
In (18), the speaker uses the distal demonstrative yule ‘that’ to signal familiarity. However, in (19) the proximal demonstrative huyu ‘this’ signals not only “larger situation” familiarity (Hawkins 1978) but also “community membership”, that is, the referent (mjukuu
wa Ndenda) lives within the speaker’s neighborhood (Clark & Marshall 1981). The use
of the distal demonstrative expression yule mjukuu wa Ndenda in (19) instead of the
proximal demonstrative expression huyu mjukuu wa Ndeda eliminates the “community
membership” implication.
3.2 Activation states
In this section, I discuss the effect of the active, semiactive and inactive activation states
on the form of the adnominal expression in the following paragraphs in turn.
As mentioned earlier, subsequent mentions of referents via anaphoric demonstrative
expressions if the referent was a continued topic were coded as active. Table 6 presents
the frequencies of the demonstrative expressions coded as active.
Table 6: Demonstrative expressions coded as active.
Proximal
Distal
Prenominal
Postnominal
Total
42
32
88
47
130
79
A few things can be said about these frequencies. First, the frequencies show that
the proximal demonstrative is used more frequently than the distal demonstrative if
the activation state of the intended referent is active (X2 (1,N=209)=12.45, p < 0.001).
Second, there is a higher frequency of proximal demonstrative in postnominal position
than in prenominal position if the activation state of the intended referent is active (X2
(1,N=130)=16.28, p < 0.001). Third, though insignificant, the frequency of the distal demonstrative in postnominal position is higher than in prenominal position if the activation
state of the intended referent is active, p > 0.05. These results corroborate the statistics
I presented on the effect of referential distance on demonstrative position of anaphoric
demonstratives in §3.1.3.
All gestural demonstratives as well as demonstratives used anaphorically after topic
shift were coded as semi-active. Table 7 and Table 8 present the frequencies of the gestural and anaphoric semi-active demonstratives.
Table 7: Semiactive gestural demonstratives.
Proximal
Distal
Prenominal
Postnominal
Total
38
2
9
3
47
5
425
Mohamed Mwamzandi
Table 8: Semiactive anaphoric demonstratives.
Proximal
Distal
Prenominal
Postnominal
Total
7
51
22
19
29
70
I have discussed the significance of the gestural demonstrative frequencies in pre and
postnominal position in §3.1.1. Here I discuss the frequencies of the anaphoric demonstratives coded as semiactive. First, the frequencies of the anaphoric semiactive demonstratives show that the distal demonstrative is used more frequently than the proximal
demonstrative (X2 (1,N=99)=16.98, p < 0.001). Second, the frequency of proximal postnominal demonstrative is higher than proximal prenominal demonstratives (X2 (1,N=29)=7.76,
p < 0.01). Third, the frequency of distal demonstrative in prenominal position is higher
than postnominal (X2 (1,N=70)=14.63, p < 0.001).
As for inactive activation state, all first mentions of familiar referents and subsequent
mentions of discourse entities after a long referential distance via adnominal demonstrative expressions were coded as inactive. Table 9 presents the frequencies of the adnominal demonstratives coded as inactive.
Table 9: Inactive adnominal demonstratives.
Proximal
Distal
Prenominal
Postnominal
Total
22
41
5
6
27
47
I have also discussed the significance of these frequencies in §3.1.4. In summary, the
prenominal position is used more frequently than postnominal if the referent is inactive.
4 Conclusion and theoretical implications
The observation that pre and postnominal demonstratives can be used as referring expressions for discourse entities has ramifications on the analysis of Swahili demonstrative expressions in pragmatics as well as syntax. In pragmatics, it has been observed
cross-linguistically that activation level of topics may be represented via different forms
of referring expressions (Gundel et al. 1993; Ariel 1988; 1991; 2001). In Swahili, the demonstratives co-occur with the noun to mark different activation levels of referents. The results of this study show that postnominal demonstratives are high accessibility markers,
prenominal demonstratives are mid-accessibility markers, and prenominal demonstratives followed by a restrictive clause are low accessibility markers.
This functional role of the demonstrative position independently motivates a syntactic
analysis of the Swahili demonstrative in pre and postnominal position (Carstens 1991;
426
25 A corpus study of the Swahili demonstrative position
2008). In the postnominal position, the unmarked order of Swahili noun modifiers is:
N>POSS>DEM>Quantifier (20) (cf. Rugemalira 2007).
(20) eneo langu
hili
lote
area 5agr.poss.1sg prox.dem all
‘all this area of mine’
Of these three types of modifiers, only the demonstrative may occur prenominally. The
functional distinction of the demonstrative in pre and postnominal position as observed
in this study rules out the possibility of these demonstratives orders being manifestation
of a single abstract syntactic structure. The different N + DEM/DEM + N constructions
correspond to different discourse needs.
Acknowledgements
I thank Dr. Laurel Stvan, Dr. Jason Kandybowicz and Dr. Jeffrey Witzel for their input on
earlier versions of this paper. I also thank the ACAL reviewers for their comments and
recommendations.
Abbreviations
Unless indicating person, numbers in glosses indicate noun class. Abbreviations follow
Leipzig Glossing Rules, with the following exceptions:
fv
intj
om
final vowel
interjection
object marker
seq
sm
stv
sequential
subject marker
stative
References
Amidu, Assibi A. 2006. Pronouns and pronominalization in Swahili grammar. Rüdiger
Köppe Verlag: Köln.
Ariel, Mira. 1988. Referring and accessibility. Journal of Linguistics 24(1). 65–87.
Ariel, Mira. 1991. The function of accessibility in a theory of grammar. Journal of Pragmatics 16(5). 443–463.
Ariel, Mira. 2001. Accessibility theory: An overview. In Ted Sanders, Joost Schilperoord
& Wilbert Spooren (eds.), Text representation: Linguistic and psycholinguistic aspects,
29–87. Amsterdam/Philadelphia: John Benjamins.
Ashton, Ethel O. 1944. Swahili grammar (including intonation). London: Longmans,
Green & CO LTD.
Auer, Peter. 1984. Referential problems in conversation. Journal of Pragmatics 8.5-6. 627–
648.
427
Mohamed Mwamzandi
Baayen, Harald R. 2008. Analyzing linguistic data: A practical introduction to statistics
using R. Cambridge University Press: New York.
Botley, Simon & Mark Anthony McEnery. 2000. Corpus based and computational approaches to discourse anaphora. Amsterdam/Philadelphia: John Benjamins.
Carstens, Vicki May. 1991. The morphology and syntax of determiner phrases in Kiswahili.
UCLA dissertation.
Carstens, Vicki May. 2008. DP in Bantu and Romance. In Cecile De Cat & Demuth Katherine (eds.), The Bantu-romance connection, 131–166. Amsterdam/Philadelphia: John Benjamins.
Chafe, Wallace L. 1987. Cognitive constraints on information flow. In Russel S. Tomlin
(ed.), Coherence and grounding in discourse. Typological studies in language, vol. II, 21–
52. Amsterdam: John Benjamins.
Clark, Herbert H. & Catherine R. Marshall. 1981. Definite reference and mutual knowledge. In Aravind K. Joshi, Bonnie L. Webber & Ivan A. Sag (eds.), Elements of discourse
understanding, 10–64. Cambridge: Cambridge University Press.
Diessel, Holger. 1999. Demonstratives: Form, function and grammaticalization. John Benjamins: Amsterdam/Philadelphia.
Fillmore, Charles J. 1975. Santa Cruz lectures on deixis, 1971. Bloomington: Indiana University Linguistics Club.
Fillmore, Charles J. 1982. Towards a descriptive framework for spatial deixis. In Robert J.
Jarvella & Wolfgang Klein (eds.), Speech, place and action: Studies in deixis and related
topics, 31–59. Chichester/New York/Brisbane/Toronto/Singapore: John Wiley & Sons
LTD.
Fillmore, Charles J. 1997. Lectures on deixis. CSLI Publications: Stanford, California.
Givón, Talmy. 1976. Topic, pronoun, and grammatical agreement. In Charles N. Li (ed.),
Subject and topic, 149–188. New York: Academic Press.
Givón, Talmy (ed.). 1983. Topic and continuity in discourse (Typological studies in language 3). Amsterdam/Philadelphia: John Benjamins.
Grosz, Barbara J., Aravind K. Joshi & Scott Weinstein. 1995. Centering: A framework for
modeling the local coherence of discourse. Computational Linguistics 21(2). 203–225.
Gundel, Jeanette K. & Thorstein Fretheim. 2006. Topic and focus. In Laurence R. Horn
& Gregory Ward (eds.), The handbook of pragmatics, 175–196. Malden, USA: Blackwell
Publishing.
Gundel, Jeanette K., Nancy Hedberg & Ron Zacharski. 1993. Cognitive status and the
form of referring expressions in discourse. Language 69(2). 274–307.
Hawkins, John A. 1978. Definiteness and indefiniteness. Groom Helm: London.
Himmelmann, Nikolaus P. 1996. Demonstratives in narrative discourse: A taxonomy of
universal uses. In Barbara Fox (ed.), Studies in anaphora, 205–254. John Benjamins:
Amsterdam/Philadelphia.
Himmelmann, Nikolaus P. 2006. Language documentation: What is it and what is it good
for? In Jost Gippert, Nikolaus P. Himmelmann & Ulrike Mose (eds.), Essentials of language documentation, 1–30. Berlin: Mouton de Gruyter.
428
25 A corpus study of the Swahili demonstrative position
Kameyama, Megumi. 1998. Intrasentential centering: A case study. In Marilyn A. Walker,
Aravind K. Joshi & Ellen F. Prince (eds.), Centering theory in discourse, 89–112. New
York: Oxford University Press.
Leonardo, Robert. 1987. Response to Wilt: Discourse distances and the Swahili demonstratives. Studies in African Linguistics 18(1). 97–105.
Mitkov, Ruslan. 1994. An intergrated model for anaphora resolution. Proceedings of the
15th International Conference on Computational Linguistics. Kyoto, Japan. In.
Mwamzandi, Mohamed. 2014. Swahili word order choices. University of Texas at Arlington. Ph.D dissertation.
Poesio, Massimo, Rosemary Stevenson, Barbara Di Eugenio & Janet Hitzeman. 2004.
Centering: A parametric theory and its instantiations. Computational Linguistics 30(3).
309–63.
Rugemalira, Josephat M. 2007. The structure of the Bantu noun phrase. SOAS Working
Papers in Linguistics 15. 135–148.
Taboada, Maite & Loreley Hadic Zabala. 2008. Deciding on units of analysis within Centering Theory. Corpus Linguistics & Linguistic Theory 4(1). 63–108.
Tamanji, Pius N. 2006. Concord and DP structure in Bafut. Afrikanistik online.
Wilt, Timothy. 1987. Discourse distances and the Swahili demonstratives. Studies in
African Linguistics 18(1). 81–95.
429
Name index
Abels, Klaus, 393, 394, 404, 405, 409
Abercrombie, David, 192
Ablorh-Odjidjah, J. R., 158
Aboh, Enoch O., 249, 406
Acker, Twanette, 58
Adams, George, 158, 159, 161
Adegbija, E, 89, 91, 93, 100
Adi, Daniel B., 158, 159
Adjei, Francisca A., 157
Ado, Derib, 66
Agyekum, Kofi, 32, 88, 91, 93–95, 97, 98,
101, 102
Aikhenvald, Alexandra Y., 325
Akinlabi, Akinbiyi, 273, 276, 277, 281
Akmajian, Adrian, 137, 138
Akpan, Ekaete E., 137, 138
Alegre, Maria, 180, 183
Alerechi, Roseline I. C., 135–138, 141,
146–148
Alsina, Alex, 301
Althusser, Louis, 22
Ameka, Felix, 157
Ameka, Felix K., 158
Amfo, Nana Aba, 157–159
Amidu, Assibi A., 414
AnderBois, Scott, 383
Anderson, Kimberley, 51
Andrzejewski, Bogumil, 210
Anthony, A., 137
Antonov, Anton, 289, 292, 293, 296, 298,
303, 304
Ariel, Mira, 413, 414, 426
Aronoff, Mark, 180
Ashton, Ethel O., 8–10, 16, 379, 383, 389,
414, 415, 421
Auer, Peter, 418
Austin, Peter K., 89–91
Awala, G., 137
Awgchew, Girma, 66
Baayen, Harald R., 176, 178, 180, 181, 422
Baayen, R. Harald, 178, 180
Baayen, Harald R., 175, 180–182, 185
Badawi, El-Said, 219–221, 223
Baier, Nico, 256, 258
Baker, Mark C., 3, 262, 272–274, 291, 301,
303, 325, 326, 337
Bakker, Peter, 6
Balsvik, Randi Ronning, 69
Bamgbose, Ayo, 90, 91, 102
Bardel, Camilla, 307, 310
Barker, Chris, 361, 370
Barlow, A. Ruffel, 8, 15
Barrett, Angeline M., 51, 52, 60
Bartsch, Renate, 361, 373
Batibo, Herman M., 129
Bauer, Laurie, 4, 7
Bax, Anna, 295
Baynham, Mike, 60
Bayram, Fatih, 29
Beck, Robert C., 60
Beck, Sigrid, 361, 362, 366
Becker-Kristal, Roy, 194, 201
Beermann, Dorothee, 368
Béjar, Susana, 265
Belletti, Adriana, 270, 279, 286
Bender, M. Lionel, 66
Bennett, Patrick R., 38
Bergvall, Victoria Lee, 394, 403
Berhanu, Kassahun, 71
Berthelette, Carol, 129
Berthelette, John, 129
Name index
Bertram, Raymond, 178
Bhat, D. N. S., 326
Bhatt, Rajesh, 361, 368, 369
Bleek, Wilhelm, 12
Blench, Rogger, 135
Bloch, Carole, 52, 55, 56
Blommaert, Jan, 24, 33
Bloom, Paul, 3
Bloomfield, Leonard, 39
Bochnak, M. Ryan, 366–368
Bogale, Berhanu, 78
Bole-Richard, Remy, 202
Bolinger, Dwight, 137
Botley, Simon, 417
Bowen, Donald, 67, 68
Bowerman, Melissa, 107
Bowern, Claire, 106, 107, 110, 112
Brenzinger, Matthias, 89, 92, 93
Bresnan, Joan, 260
Broadbent, Donald E., 178
Brown, Dunstan, 222
Bryant, Peter E., 55
Buell, Leston, 298
Bulcha, Mekuria, 66
Byarushengo, Ernest Rugwa, 7, 8, 10, 12
Bybee, Joan L., 175, 180–182, 184
Cable, Seth, 377, 378, 381, 387–389
Caesar, Regina O., 157–159
Cain, Kate, 60
Cameron, Deborah, 22
Cammenga, Jelle, 393, 394
Canonici, Noverino N., 53, 55
Carrington, John F, 10
Carstens, Vicki May, 414, 421, 426
Castroviejo, Elena, 373
Cenoz, Jasone, 310
Chafe, Wallace L., 108, 325, 326, 335, 337,
413, 414, 419
Chelliah, Shobhana, 105, 106
Chen, Lily, 31
Chiswick, Barry, 79
Chomsky, Noam, 266, 276, 279, 308, 364,
402
432
Clark, Herbert H., 421, 425
Cleghorn, Ailie, 52
Cobarrubias, Juan, 66, 73
Collins, Chris, 246, 249
Comfort, Jade, 206, 209
Comrie, Bernard, 6, 377
Cook, Thomas L., 276
Cooper, Robert L., 66, 68
Corbett, Greville G., 206
Coulmas, Florian, 97
Cover, Rebecca, 235
Craig, Colette Grinevald, 97
Crass, Joachim, 341, 342, 346, 350, 352,
354–356
Creissels, Denis, 292
Cresswell, Max J., 362, 363, 370
Croft, William, 346
Crowhurst, Megan, 8
Crystal, David, 97, 143
Dahl, Östen, 4, 377
Danti, Alexis, 157
David, Stephen O., 138
De Dreu, Merijn, 298
De Jong IV, Nivja H., 178
de Reuse, Willem, 105, 106
Debra, Spitulnik, 94
Del Prado Martín, Fermín Moscoso, 180,
181
den Dikken, Marcel, 285
Derhemi, Eda, 89, 94, 97, 102
Diercks, Michael, 294, 295, 301
Diessel, Holger, 352, 356, 414, 417
Dimitriadis, Alexis, 177
Dixon, R. M. W., 158, 165, 168–170, 325,
336, 342, 345
Donwa-Ifode, Shirley O., 136, 138, 139
Duncan, Lynne G., 55
Dyakanova, Marina, 406
Dzameshie, Alex, 157
Egan, Kieran, 55
Ekwulo, S. A., 136, 138, 139
Engstrand, Olle, 9
Name index
Eska, Joseph F., 40
Eslinger, Adele, 394
Essien, Okon E., 275, 276
Evans, Cay, 51
Evans, Rinelle, 52
Evans, Roger, 223–226
Everest, Alton F., 198
Fairclough, Norman, 23–25, 34
Falk, Yehuda, 307, 310
Fara, Delia Graff, 382
Faraclas, Nicholas, 136
Faris, David, 345
Faye, Souleymane, 254
Ferrand, Carole T., 198
Fillmore, Charles J., 413, 414
Fine, Kit, 382
Fishman, Joshua A., 73, 99
Flynn, Suzanne, 307–310
Foote, Rebecca, 181
Fowler, Roger, 26
Fretheim, Thorstein, 414
Fromkin, Victoria, 136, 137, 192, 202
Fry, Dennis Butler, 200
Gadalla, Hassan AH, 219, 220
García-Mayo, Maria P., 310
Garret, Paul B., 88
Garrett, Andrew, 40
Gazdar, Gerald, 223–226
Gebreselassie, Fisseha, 79
Gee, James Paul, 56
Getachew, Anteneh, 66
Giannakidou, Anastasia, 285
Givón, Talmy, 3, 4, 415, 418, 421
Glenn, Christine G., 51, 53, 58
Gordon, Peter, 180, 183
Gore, Peter A., 8, 10
Goswami, Ursula, 55
Gramsci, Antonio, 25
Greenberg, Joseph H., 38, 175
Griffith, Penny L., 53
Grin, Francois, 79
Grinsell, Timothy, 382
Grosz, Barbara J., 418, 420
Gudina, Merera, 69
Gundel, Jeanette K., 413, 414, 426
Guthrie, Malcom, 8, 10, 12, 13
Gutierrez-Clellen, Vera F., 56
Hagiwara, Hiroko, 181
Halliday, Michael, 31
Halpert, Claire, 11, 294, 299, 301, 302
Hammond, Michael, 221
Hara, Agness, 51, 57–59
Harackiewicz, Judith M., 60
Hardré, Patricia L., 60
Harford Perez, Carolyn, 301
Harizanov, Boris, 253, 254, 264
Hartmann, Katharina, 393, 396, 404
Harves, Stephanie, 289–291, 293–295,
298, 301–304
Hasty, Jennifer, 24
Haviland, John Beard, 106, 349
Hawkins, John A., 425
Hay, Jennifer, 4, 7, 175, 178, 180–182, 185
Hayashi, Midori, 377
Hayward, Richard J., 206
Heath, Shirley Brice, 56
Hébert, Jean R. P, 130
Heim, Irene, 361, 363, 370, 387
Heine, Bernd, 4, 176
Henderson, Robert, 383
Hérault, Georges, 202
Hester, Eva Jackson, 56
Heugh, Kathleen, 66, 68, 70
Hidi, Suzanne, 60
Hill, Deborah, 106
Himmelmann, Nikolaus P, 97–99, 414,
418, 424
Hinds, Martin, 219–221, 223
Hippisley, Andrew, 222
Holmberg, Anders, 273
Holmes, Janet, 92
Hopper, Paul J., 343, 346, 354
Horn, Nancy, 67, 68
Horvath, Julia, 402, 403
Hovdaugen, Even, 337
433
Name index
Howell, Anna, 361, 362, 366
Hübschmann, Heinrich, 40
Hufeisen, Britta, 310
Hutchison, John P, 355
Hyman, Larry M., 7–11, 142, 378
Hyter, Yvette D., 56
Jacques, Guillaume, 289, 292, 293, 296,
298, 303, 304
Jaffe, Alexandra, 93, 99
Jakobi, Angelika, 341, 342, 345, 346, 349,
350, 352–356
Jarvis, Scott, 313
Jerro, Kyle, 12
Jessner, Ulrike, 310
Johnson, Keith, 191, 200
Kameyama, Megumi, 418
Kamp, Hans, 382
Katamba, Francis X., 11, 178, 179
Kaye, Jonathan D., 39–41, 44, 48
Kayne, Richard S., 235, 236, 247, 248,
282, 289–291, 293–295, 298,
301–304
Kennedy, Christopher, 361–363, 366,
370, 372, 373, 382, 386, 388
Khidir, Zakaria Fadoul, 349, 356
Kihm, Alain, 219–222
Kimenyi, Alexandre, 12
Kiparsky, Carol, 241
Kiparsky, Paul, 241
Kirwan, Brian Edmond Renshaw, 8, 10
Kiss, Katalin É., 270, 282, 286
Kisubika-Musoke, E. M., 379
Klein, Wolfgang, 387, 389
Koffi, Ettien N., 192, 195, 200, 201
König, Christa, 341, 342, 346, 355
Koopman, Hilda, 39–41, 43, 45, 46
Kouadio, Jérémie N., 202
Kramer, Ruth, 253, 254, 264
Kratzer, Angelika, 377, 387
Krifka, Manfred, 11, 270, 284, 286, 396
Kroger, Rolf O, 24
434
Kropp Dakubu, Mary Esther, 158, 159,
161
Kusters, Wouter, 4, 5, 16
Kuteva, Tania, 4
Labarba, Richard C., 137
Labov, William, 56, 200, 202
Ladefoged, Peter, 191, 196, 197, 200, 201
Lambrecht, Knud, 341, 343, 344
Lasnik, Howard, 276
Le Moal, Guy, 130
Lecarme, Jacqueline, 208
Lechner, Winfried, 369
Lehtonen, Minna, 181
Leonardo, Robert, 424
Leow, Ronald P., 321
Lewis, Huw, 90, 96, 99
Lewis, M. Paul, 37–39, 47, 219, 237
Leyew, Zelealem, 66
Lin, Jo-wang, 369
Lodhi, Abdulaziz Y, 9
Lüpke, Friederike, 93, 129
Maddieson, Ian, 7, 8, 10, 192
Makgamatha, Phaka Moffat, 53, 55
Mandler, Jean M., 51, 53
Manessy, Gabriel, 117, 125, 126
Marchese Zogbo, Lynell, 38, 40, 42
Marchese, Lynell, 37–41
Marfo, Samuel, 21
Marlo, Michael R., 410
Marshall, Catherine R., 421, 425
Marten, Lutz, 295
Matthewson, Lisa, 115
Matushansky, Ora, 253, 264
Mayer, Mercer, 57
McCarthy, John J., 221, 222
McCloskey, James, 258
McEnery, Mark Anthony, 417
McGregor, William B., 341, 342, 348, 356
Mchombo, Sam, 329
McLaughlin, Fiona, 240, 254
McNab, Christine, 66, 68, 70
McNally, Louise, 362, 363, 370
Name index
McWhorter, John, 3–5, 8
Meek, Margaret, 55
Meeussen, Achille Emile, 10, 13
Mekonen, Walelign, 69
Melchert, H. Craig, 40
Menn, Lise, 137
Mensah, Emanuel N. A, 201
Merchant, Jason, 369
Meyer, Michael, 24, 26
Michaels, Sarah, 56
Miehe, Gudrun, 121
Miestamo, Matti, 3
Mikkelsen, Line, 341, 343, 354
Miler, Zdenek, 57
Mithun, Marianne, 345, 377
Mitkov, Ruslan, 415
Mohamed, Mohamed Abdulla, 176, 177,
179
Moltmann, Friederike, 362, 370, 373
Moriarty, Máiréad, 94
Mosel, Ulrike, 106, 108
Moshi, Lioba, 260
Motshwari, Julia Jakoentle, 53, 55
Mous, Maarten, 87, 88, 90, 91, 93, 209
Mucha, Anne, 377
Mufwene, Salikoko S., 15, 88, 93, 99, 129,
130
Mullis, Ina V. S., 60
Muriungi, Peter, 393, 394, 404, 405, 409
Mutonyi, Nasiombe, 332
Mwamzandi, Mohamed, 424
Mwita, Chacha, 393, 394, 403, 407, 409,
410
Myers, Scott, 8
Naden, Anthony J., 119, 121, 122, 125, 129
Naden, Tony, 157
Naess, Åshild, 337
Nahir, Moshe, 96
Ndayiragije, Juvénal, 301
Nettle, Daniel, 119, 129
Nguako, Nduvaa Erna, 292
Nichols, Johanna, 3, 4, 7
Nicolas, David, 362, 370
Ninsin, Kwame, 22
Nkosana, Leonard B. M., 52
Nomlomo, Sylvia Vuyokazi, 52
Ntuli, Cynthia D., 55
Nurse, Derek, 13, 377
Oakhill, Jane V., 60
Obeng, G. Samuel, 65
Obiechina, Emmanuel, 53, 55, 56, 59
Odonkor, Padi, 158, 159
Ojukwu, Obed, 137, 138
Oomen, Antoinette, 209
Orman, Jon, 52
Osam, Emmanuel Kweku, 157
Otoo, Yvonne A., 157, 158
Owusu, William, 21
O’Grady, William, 136, 137
O’Neill, Daniela K., 60
O’Neill, Robert, 315
O’Shannessy, Carmel, 88
Pandharipande, Rajeshwari V, 89
Pankhurst, Richard, 67
Paster, Mary, 205, 206, 209, 210, 217
Patten, Allan, 73, 75
Payne, Thomas Edward, 342
Pederson, Eric, 107
Perlmutter, David M., 276
Pesetsky, David, 285
Philips, Susan U, 92
Philipson, David, 66
Phillipson, Robert, 87, 88, 90
Pinker, Steven, 3
Plag, Ingo, 180
Platzack, Christer, 273
Poesio, Massimo, 418
Pohlmann, Ken C., 198
Pokuaa, Christiana, 157
Pollard, Carl, 222
Pollock, Jean-Yves, 273
Prah, Kwesi Kwaa, 100
Preminger, Omer, 265, 266
Price, Glanville, 119, 129
Prince, Alan S., 221, 222
435
Name index
Prince, Ellen, 341, 343, 344, 348
Purvis, M. Tristan, 65
Quaireau, André, 195
Ratcliffe, Robert R., 221, 222
Reichenbach, Hans, 387
Reisigl, Martin, 26
Renaudier, Marie, 253, 254
Retord, Georges L. A., 195
Rezac, Milan, 265
Richardson, John, 25, 26, 29, 30
Riedel, Kristina, 295
Ring, J. Andrew, 110
Ringbom, Hakan, 309
Ringe, Don, 40
Rizzi, Luigi, 270, 275, 282, 284, 401, 403
Robinson, Peter, 321
Romaine, Suzanne, 119, 129
Rooth, Mats, 396, 401
Ross, John R., 283
Rothman, Jason, 308, 310
Rothman, Jason, 308, 310
Rowlett, Paul, 308
Rugemalira, Josephat M., 427
Rumsey, Alan, 349, 350, 354, 356
Russell, Joan, 12, 13
Ryan, Kevin, 198
Saeed, John I., 206–208, 216
Sag, Ivan A., 222
Salffner, Sophie, 98
Sallabank, Julia, 89–91
Salzmann, Zdeněk, 92
Sampson, Geoffrey, 3
Saville-Troike, Muriel, 92
Schadeberg, Thilo C., 10, 12
Schiffman, Harold, 73
Schmidt, Richard, 321
Schultink, Henk, 180
Schunk, Dale H., 60
Schwager, Magdalena, 373
Schwartz, Jean-Luc, 194, 201
436
Schwarz, Florian, 393, 394, 396, 398,
401–403, 406, 409, 410
Seidl, Amanda, 177
Selkirk, Elisabeth, 398
Sharman, J. C., 378
Shibatani, Masayoshi, 3
Showalter, Stuart D., 129
Shuman, Amy, 56
Smith, Lahra, 65, 66, 68
Smith, Michèle, 60
Soames, Scott, 382
Souto-Manning, Mariana, 51, 56
Stalin, Joseph, 69
Stein, Nancy L., 51, 53, 58
Sterk, Jan P, 38
Stoddart, John, 78
Storch, Anne, 93, 129
Swadesh, Morris, 105, 106
Sylla, Yèro, 238, 240
Taboada, Maite, 418
Taft, Marcus, 178, 180
Takahashi, Shoichi, 361, 368, 369
Tamanji, Pius N., 414
Tamba, Khady, 246, 248, 249
Tappe, Heike, 51, 54, 55, 57, 58
Thomas, Erik R., 192, 197
Thompson, Sandra A., 343, 346, 354
Tikly, Leon, 51, 52
Torrence, Harold, 246, 248, 249
Tosco, Mauro, 209, 216
Triebel, Armin, 52
Trilsbeek, Paul, 97
Trudgill, Peter, 3, 4, 6, 15
Tzadua, Paulos, 66
Udoinyang, Mfon. E, 275
Urua, Eno E., 273, 276, 277, 281
van Dijk, Teun, 22–25, 33, 34
Vaughan, Sarah, 71
Vaux, Bert, 106
Vennemann, Theo, 361, 373
Verstraete, Jean-Christophe, 342
Name index
Vogler, Pierre, 38
Von Stechow, Arnim, 362
Wagner, Daniel A., 52
Waletzky, Joshua, 56
Weber, David J, 326
Welmers, William E., 38, 191, 192
Westby, Carol E., 56
Whaley, Lindsay J, 326
Williamson, Kay, 135, 136
Williamson, Timothy, 382
Willie, Willie U., 272–275
Wilt, Timothy, 424
Winkelmann, Kerstin, 118, 120, 121, 123,
130
Wittenburg, Peter, 97
Wodajo, Mesfin, 79
Wodak, Ruth, 23–26, 29
Wolff, H. Ekkehard, 210
Wood, Linda A, 24
Woodbury, Anthony, 98
Yiannis, Gabriel, 23
Yoon, Tae-Jin, 198
Yul-Ifode, Shirley, 137, 147
Zabala, Loreley Hadic, 418
Zeller, Jochen, 295
Zemelak, Mehari, 66
Zerbian, Sabine, 11
Zewde, Bahru, 67, 69, 71
Zimmermann, Malte, 393, 396, 404
Zogbo, Raymond Gnoleba, 39–41
437
Language index
Adele, 410
Afar, 68, 77, 206
Afrikaans, 58
Akan, 28, 32, 33, 92, 927 , 93, 939 , 95, 170,
191, 195, 201, 202
Akpo, 136, 141–146, 149, 153
Akwa, 272
Aluu, 136, 141, 143–146, 149, 153
Amharic, 66, 68–71, 74, 75, 7936 , 264, 290
Animere, 110
Anyi, 191, 192, 194–198, 200–202
Arabic, 4, 8, 9, 13, 15, 16, 88, 91, 92, 927 ,
119, 214, 219, 220, 2201 , 221–
223, 2236 , 230, 290, 293, 304,
344, 346
Armenian, 40, 290
Asante, 91, 92
Ashanti, 33, 91
Ayacucho, 290, 304
Bambara, 129
Baule, 202
Beria, 342, 344, 345, 3452 , 346, 349, 350,
352–354, 3544 , 355, 356
Bobo, 117, 118, 128–130
Bulgarian, 264, 265, 290
Bunuba, 350, 356
Burunge, 210
Bété, 37–48
Chichewa, 57–59, 329
Chinese, 3695
Cree, 6
Dangme, 157–171
Day, 110, 111, 113–115
Dazaga, 355, 357
Deg, 364
Dem, 422
Dida, 37–40, 43, 44, 47, 48
Dogon, 131
Dutch, 181, 215, 290
English, 57, 573 , 58, 67, 68, 683 , 71, 78,
7936 , 88, 90, 903 , 904 , 91, 915 ,
927 , 96, 99, 100, 126, 129, 165,
170, 1751 , 176–178, 180, 181, 188,
200–202, 212, 214, 220, 277,
28212 , 290, 2933 , 301, 308, 311–
315, 326, 327, 335–338, 344,
352, 362, 363, 367, 368, 3683 ,
3684 , 369–371, 3792 , 382, 386–
388, 39810 , 401, 40322 , 40928 ,
421
Ewe, 91–93, 98, 158, 170, 290, 304
Faroese, 5
Finnish, 181, 293, 294, 298
French, 6, 39, 67, 68, 683 , 88, 91, 96, 129,
181, 196, 2735 , 290, 308, 311–
314, 3451 , 349, 355
Ga, 157–171
Gbadi, 38–48
Geez, 66, 696
German, 96, 118, 290
Gikuyu, 7, 9, 105 , 127 , 14, 15
Greek, 290
Gùrùntùm, 396, 404
Hausa, 92, 93, 326
Haya, 7–10, 105 , 11–13, 138 , 15
Hindi, 290, 368
Hungarian, 290, 40423
Language index
Ibibio, 269–272, 2721 , 273, 2735 , 274, 276,
277, 279–281, 28211 , 283–287
Igbo, 93, 135
Igboid, 135, 136
Ikwere, 135, 136, 138, 140–142, 145–148,
154
Inuktitut, 377
Italian, 67, 212, 214, 2799 , 284, 290
Japanese, 181
Jula, 93, 118, 124–127, 129–131
Juma, 417
Kan, 28
Kana, 135
Kanuri, 355
Kiitharaka, 394, 404, 40524 , 409
Kikuyu, 93, 394, 396, 3966 , 39811 , 401,
40219 , 403, 40322 , 406, 409,
410
Kim, 361, 363, 364, 369
Kinande, 12
Kinyarwanda, 7–10, 105 , 11, 12, 126 , 13–15
Kiswahili, 91, 329
Kongo, 9
Krobou, 201, 202
Kru, 37–42, 44, 47, 48
Kube, 342, 344, 352, 356, 357
Kuria, 290, 292, 2933 , 2945 , 303, 393,
394, 396, 39810 , 39811 , 401, 403,
40322 , 40323 , 404, 40423 , 406–
410
Lakota, 37–40, 44, 48
Latin, 87, 88, 290
Lingala, 7–10, 105 , 11–13, 138 , 15, 93, 308,
311–314
Lobi, 38, 129, 130
Lubukusu, 2627 , 292, 301, 325, 327, 328,
3284 , 329, 330, 332–338
Luganda, 7–10, 105 , 11, 127 , 14, 15, 362,
363, 365, 366, 368, 3684 , 369–
371, 373, 374, 377–379, 3792 ,
3793 , 380, 382–384, 386–389
440
Maay, 205–207, 209, 210, 2103 , 211, 213–
217
Mali, 93, 129, 131
Manipuri, 326, 327
Maya, 290
Mayali, 326, 327
Medumba, 377
Michif, 6
Mohawk, 290, 326, 327, 337, 338
Mooré, 117
Ndanda, 424
Norse, 5
Nouchi, 129
Omo, 206
Omuanwa, 136, 141, 142, 149, 153
Oromo, 66, 662 , 67, 69, 71, 74, 7417 , 75, 76,
206, 210
Peru Quechua, 326
Portuguese, 88, 91, 290
Pulaar, 235, 2351 , 236–240, 242–244,
246–250
Quechua, 290, 293, 304, 326, 327
Rendille, 206, 209, 217
Seereer, 253, 254, 2542 , 255, 256, 258–
260, 264–267
Seneca, 326, 327, 3377
Setswana, 91, 290, 292, 2922
Shona, 292
Somali, 68, 69, 93, 206–208, 212, 216, 217
Spanish, 181, 212, 290
Swahili, 4, 6, 7, 71 , 8, 9, 94 , 10, 105 , 11–
16, 1751 , 176–178, 181–183, 185,
18511 , 188, 212, 290, 292, 2921 ,
293, 2933 , 294, 2945 , 295–299,
301–303, 413–416, 421, 424,
426, 427
Swedish, 683 , 290
Tama, 354
Language index
Teda, 355
Terik, 6
Tiefo, 117, 1172 , 118, 1184 , 119–131
Tooro, 237
Tso, 169
Tunni, 206, 209, 216, 217
Twi, 91, 92, 98
Vata, 38–48
Viemo, 129
Waru, 350, 356
Wolof, 93, 129
Xhosa, 12, 57
Zigua, 212, 213
Zulu, 9, 11, 57, 290, 292, 2921 , 293, 2933 ,
294, 2945 , 295–300, 3007 , 301–
303
441
Subject index
adult language, 146–148
adult speech, 137, 145, 147
affricate, 119, 410
affrication, 144, 147
agentive, 95, 178, 182–186, 188, 346, 349,
352, 357
agree, 57, 109, 191, 220, 246, 247, 278, 309,
326, 330–332, 413
agreeing complementizer, 243
alveolar, 9, 136, 137, 142, 144, 145, 410
applicative, 126 , 16, 47, 259–262
archival database, 98
Botswana, 91
Burundi, 6, 7
Cameroon, 237
causative, 43–47, 259–262, 266, 2747
causatives, 12, 258, 261, 262, 266, 2733
causee, 261, 266
child language, 136, 138, 142, 146–148,
154
child language acquisition, 136, 147
clausal determiner, 243
clause, 11, 29, 123, 162, 165, 166, 171, 235,
241, 243, 245, 246, 2563 , 258,
259, 261, 266, 270, 272, 2733 ,
2747 , 284, 301, 342, 343, 346,
348, 351, 352, 357, 394, 396,
400, 403, 40322 , 408, 409, 417,
418, 426
clitic, 296
clitic doubling, 264, 265
complementary distribution, 256, 276,
345, 407, 409, 410
complementizer, 235, 2351 , 243–247,
249, 250, 275, 276, 282, 28211 ,
283, 287
complex head, 264, 265, 273, 275
compound, 95, 3793
Congo, 6, 7, 16, 37, 38, 93, 121, 135, 157,
176, 237, 272, 314, 315
consonant mutation, 240, 244–246
consonant substitution, 137, 138
cultural heritage, 95, 100
Darfur, 342, 344
decolonisation, 92–96, 99–101
denasalization, 137, 148
devoicing, 137, 147
distal, 352, 413–416, 420–426
ditransitive, 259–263, 394, 415
dominant language, 57, 68, 77, 87, 89, 93,
97
double object, 259–262, 266
dual, 176, 186
Egypt, 219
endangered language, 89, 100, 106
endangerment, 89, 94, 96, 97, 100, 102
Eritrea, 67, 96, 9612
Ethiopia, 66, 661 , 662 , 67, 68, 683 , 69–71,
718 , 73, 7417 , 75, 78, 79
extraction, 256, 2563 , 257, 276, 28212 ,
295–298, 300, 415
factive, 235–237, 241–243, 245, 2455 ,
246–248, 250
Focus, 238
focus, 2351 , 238, 2456 , 2467 , 256, 264,
269–271, 2735 , 275–279, 2799 ,
280–282, 28210 , 28211 , 28212 ,
Subject index
283–286, 28614 , 287, 342–346,
3463 , 347–353, 356, 393, 394,
3943 , 395, 396, 398, 39810 , 399,
39912 , 39913 , 400, 40014 , 40015 ,
40016 , 401, 40219 , 403, 40323 ,
404, 40423 , 405–409, 40928 ,
410
fricative, 9, 119, 137, 142, 145, 328, 410
fricatives, 9, 119, 136, 139, 142–147, 154,
328
functional load, 89, 97, 102
Ghana, 21–24, 26–28, 30, 32–34, 89–91,
915 , 92, 93, 939 , 94, 95, 9612 , 98,
99, 106, 110, 117, 158, 171
glottal, 119, 120, 145, 154, 223, 224
glottis, 142
Guinea, 90, 237, 350, 356
head movement, 254, 264–266, 270, 271,
275, 276, 284
hierarchy, 226, 227, 274
implosive, 10
implosives, 10, 119, 136, 142, 154
incorporation, 262–264, 266, 291, 302
indigenous language, 52, 66, 68, 77, 89
infinitival, 249, 250, 2509 , 406, 407
infinitive, 235, 238, 239, 245, 246, 249,
362, 366
Kenya, 6, 7, 915 , 93, 212, 325, 393
labial implosive, 145, 147
labial plosive, 145, 147
labialized, 142, 147
language acquisition, 4, 94, 143, 308, 309,
318
language attrition, 93
language contact, 4, 6, 15, 16, 88, 93, 119,
124, 126, 130
language death, 89, 94, 97, 119
language decay, 89
language decolonization, 94, 97
444
language development, 99, 100
language documentation, 94, 97, 98, 101,
105, 107, 112, 114
language endangerment, 78, 88, 89, 92,
100–102
language imperialism, 88, 96, 101
language maintenance, 94, 96, 97
language planning, 101
language policy, 70, 91, 9613 , 99, 101
language preservation, 97, 99
language resistance, 96
language revitalisation, 89
language revitalization, 94, 96
language shift, 88, 882 , 89, 90, 93, 94, 96,
97, 100–102
language survival, 89, 102
left dislocation, 258
Lesotho, 552
lexicology, 100
Libya, 219
Likert scale, 57
linear order, 259
lingua franca, 4–6, 16, 90, 903 , 93, 129
linguistic dominance, 88
linguistic imperialism, 87, 88, 881 , 89, 92,
94, 100–102
linguistic pluralism, 94, 96, 9612
literacy, 52, 55, 56, 59, 69, 697 , 94, 196
major, 71 , 22, 37, 38, 44, 51, 59, 607 , 67, 75,
77, 87, 89, 90, 93, 95, 9613 , 97,
98, 109, 113, 114, 119, 135, 137,
202, 223, 327
Malawi, 54, 552 , 56, 57, 59
minority language, 88, 99, 126
movement, 47, 67, 69, 71, 236, 248, 250,
2563 , 257, 258, 264, 265, 270,
272, 275, 276, 278, 281, 28210 ,
283–285, 364, 402, 40220 ,
40321 , 40423
multilingualism, 6, 75, 76, 78, 96, 130, 310
nasalization, 148
Subject index
negation, 13, 15, 47, 272, 273, 2733 , 274,
2747 , 287, 326, 409, 40928 , 410
Niger, 93, 117, 157
Nigeria, 90, 915 , 93, 117, 135, 136, 272
nominalization, 176–178, 182, 183, 241,
367, 368, 370, 374, 407
nominalize, 295
noun, 10–12, 30, 121–123, 161–163, 165–
167, 169–171, 176, 178, 182, 184,
1849 , 188, 207, 212–216, 2167 ,
217, 220, 224, 225, 237–239,
241, 243, 245–247, 251, 301,
327, 329–335, 345, 349, 350,
352, 355, 356, 3707 , 371, 404,
413, 415, 417, 426, 427
noun class, 10, 121, 122, 176–178, 182–184,
1849 , 185, 18510 , 188, 2351 , 238,
239, 246, 247, 249, 250, 299,
339, 362, 401, 410, 413, 427
oral literature, 95, 97, 98, 100
orthography, 1962 , 3451 , 3452
palatal, 119, 410
passive, 12, 13, 258, 263, 264, 266, 343,
349, 352
perfect, 44, 192, 308, 311, 313, 315, 316,
318, 319
perfective, 15, 44, 123–125, 287, 3452
pharyngeal, 119
phonological development, 136
phylum, 135, 210
plosive, 142
policy, 66, 67, 69, 70, 72, 76–79, 94, 96
postnominal, 413, 414, 416, 420–427
predicate, 168, 170, 262, 302, 336, 338,
343, 344, 350, 357, 362, 363,
366, 373, 382, 386, 406, 40625
preposition, 292, 297, 300, 302, 333, 334
probe, 264–266, 2668
progressive, 123, 125, 315
pronoun, 123, 124, 2351 , 243, 246, 247,
257–266, 2668 , 297, 357, 387,
388
pronoun incorporation, 254
proverb, 95
recontextualisation, 95
relative clause, 236, 247, 248, 28210 , 329,
343, 350, 351, 40322
relativization, 248
relativizer, 243, 245, 246, 352, 357
resumptive, 257, 258, 296, 297
root, 43, 160, 170, 1827 , 18510 , 221–223,
225, 229, 230, 238, 246, 2721 ,
2736 , 277, 328–330, 332, 409
Rwanda, 6, 7, 11
second language, 6, 7, 16, 52, 90, 96, 99,
129, 181
Senegal, 93, 235, 237, 253
social stratification, 92, 99
Somalia, 206
sorites paradox, 382, 384
specifier, 264, 275, 40321
spirantization, 136, 146
subject agreement, 206, 240, 272, 3283 ,
335, 339
substitution, 136–138, 141–147
successive affrication, 147
Sudan, 237
suffixation, 160, 164, 254, 258–260, 2606 ,
261–266, 329
Tanzania, 6, 7, 91, 393
temporal remoteness, 377, 378, 3781 ,
3793
Togo, 93, 117, 158
vagueness, 378, 381, 382, 387, 38711 , 388,
38813 , 389
velar, 9, 136, 137, 142, 160, 410
verb, 3, 5, 12, 13, 31, 44–46, 124, 125,
159, 160, 168–171, 177, 1772 , 178,
1785 , 188, 2091 , 235, 238, 240,
2404 , 241, 242, 244–247, 249,
250, 2509 , 251, 253–256, 2563 ,
257–266, 269–271, 2721 , 2732 ,
445
Subject index
2735 , 2736 , 274, 2747 , 275–281,
28210 , 28212 , 283, 28313 , 284–
286, 289–292, 2922 , 293, 2933 ,
294–299, 301, 302, 315, 319–
321, 328, 329, 336, 338, 341,
349, 351, 357, 362, 365–368,
3683 , 369, 370, 389, 3944 , 39811 ,
406, 407, 409, 40927 , 417
voice, 29, 30, 34, 697 , 9916 , 263, 266, 352,
3545 , 357
word order, 44, 237, 244, 246, 247, 250,
260, 261, 271, 272, 310, 334, 350,
351, 394, 395, 413
Yemen, 219
446
Did you like this
book?
This book was brought to you for
free
Please help us in providing free access
to linguistic research worldwide. Visit
http://www.langsci-press.org/donate to
provide financial support or register as
a community proofreader or typesetter
at http://www.langsci-press.org/register.
language
science
press
African linguistics on the prairie
African linguistics on the prairie features select revised peer-reviewed papers from the 45th Annual Conference on African Linguistics, held at the University of Kansas. The articles in this
volume reflect the enormous diversity of African languages, as they focus on languages from all
of the major African language phyla. The articles here also reflect the many different research
perspectives that frame the work of linguists in the Association for Contemporary African Linguistics. The diversity of views presented in this volume are thus indicative of the vitality of current African linguistics research. The work presented in this volume represents both descriptive
and theoretical methodologies and covers fields ranging from phonetics, phonology, morphology, typology, syntax, and semantics to sociolinguistics, discourse analysis, language acquisition,
computational linguistics and beyond. This broad scope and the quality of the articles contained
within holds out the promise of continued advancement in linguistic research on African languages.